Long story short. Today I got pretty bored. I’ve been wanting to put my 1090T under dry ice for quite some time now. However, my bench partner has been busy with work so I could not obtain the dry ice pot.

So I was thinking of ways I could cool down my processor, in order to play around with my new 6 core. Since I own a Corsair H50 AIO Watercooling kit, I decided to have some fun.

I hung the radiator in an old lunch cooler and filled it to the brim with ice. I also threw a bit of water in for more cooling surface area. I figured this could get my processor 5°C or below. I was right. Yes, the slushbox idea is not a brand new innovative one, but I wanted to try my chip under something else then regular cooling for now.

So I pushed my processor to frequencies I only previously thought could be attained on dry ice (at least with my deneb’s). It seems these new Thuban chips pack quite the punch when it comes to running up the frequency.

Anyways. Enough with the chit chat. I threw together another youtube short just like the good old days when I played with my denebs. So here you are: http://www.youtube.com/watch?v=Wwb1K3_G7Mg

[youtube=http://www.youtube.com/watch?v=Wwb1K3_G7Mg]

Here are all the screenshot results and validations:

4250MHz Prime 95 Stable 1 Hour

I wanted to see how the slushbox worked under a real stress and if the chip would stay stable. The slushbox doesn’t deal with loads too well, but is still kept below 32C at all times. I believe if I had a good enough water setup, I could attain this stable speed no problem without any slushbox at all.

SuperPi 1m @ 4662MHz

I ramped up SuperPi to see what this chip can do. On my previous deneb chips, SuperPi 1m was my bread and butter, my favourite benchmark to run. Since SuperPi is a single threaded application, I imagine if I disabled some cores or ramped up only one core I could get better results.

SuperPi 32m @ 4500MHz                                                        SuperPi 32m @ 4571MHz

My first run was kind of on the safe side. To avoid a crash. However for second run I ramped up the processor a bit. This was the max I could go while still using all 6 cores. Once again, being single threaded, disabling cores might be able to help my clocking.

wPrime 32m @ 4500MHz

wPrime 1024m @ 4250MHz                                    wPrime  1024M @ 4409MHz

PiFast @ 4500MHz PiFast @ 4553MHz

Cinebench 11.5 @ 4250MHz Cinebench 11.5 @ 4427MHz

3DMark Vantage @ 4427MHz

SiSoft Sandra @ 4427MHz

CPUZ SS @ 4730MHz – Validation

In the video I attained a speed of 4.75GHz, however it was not stable for validation nor screenshot. It was only a taken as a video clip. The cpu did not last very long at 4.75GHz and crashed soon after.

Upon plugging the key in, I was greeted by a series of statements to complete and the option to generate a passcode. After completing the statements, I generated a passcode, removed the key from the computer and opened the phase 2 envelop. Phase 2 required placing the key with the passcode into the USB slot on the large case, pressing a button on the case and hoping I generated the correct passcode.

Luckily, I did generate the correct passcode and the case was now unlocked. When I opened the case, I could not believe what was inside: an Intel Core i7-875k, an IntelDP55SB Extreme Series, and two 4Gb kits of Patriot Viper II Series, Sector 5 Edition PC3-19200 (2400Mhz)

That brings us to phase three of the mission, using this unlocked version of Intel’s ultimate smart performer, Core i7.

The setup:

i7-875k
8GB PC3-19200*
Intel DP55SB
EVGA GTX 260 c216 (600/1458/1000, core/sp/mem)*
160GB Western Digital Caviar SE *OCZ ModXstream 700w*
Cooler Master Hyper 212+ with two Cooler Master R4-C2R-20AC-GP in push/pull configuration.
Win7 Pro x64*

*For comparison, I used a Core i7 920 (2.66Ghz) on an EVGA 3x SLI (E758) board along with these components.

Although a vital part of any computer, the CPU heat sink was not included in the case. I opted for this setup for two reasons, the first being that I could run out to MicroCenter and pick up the Hyper 212+ and two extra high CFM, low DB fans for a total of $30, and the second being that the Hyper 212+ cools like a beast. Originally, I had planned on liquid cooling, but after I got an 1156 retention bracket for my block, my pump died so I stuck with this setup throughout my testing.

A video overview of the unlocking portion of the mission:

Before taking a look at the performance, let’s look at the features of the i7-875k. The i7-875k sports 4 physical cores, 8 logical via Intel Hyper-Threading @ 2.93Ghz, 3.60Ghz with Turbo Boost enabled and 8Mb Smart Cache. Intel Turbo Boost Technology, which can be controlled in BIOS, dynamically changes processor frequency when working under specified power and thermal limits in order to provide a smoother computing experience. Hyper-threading allows you to run a up to 8 threads simultaneously increasing your productivity and the integrated memory controller supports dual channel DDR3 memory. This gem sure packs a powerhouse in a tiny package.

The “K” in the model number indicates that the processor is unlocked. What this means is that the processor has unlocked Turbo Boost multipliers, allowing for quick and easy overclocking as well as unlocked voltage and memory frequency. Pairing an unlocked processor with the Intel DP55Sb motherboard allows for full control of your computing experience. The highly tunable BIOS let you easily take advantage of unlocked multipliers, and overclock like mad.

On to the stressing, benching and overclocking.

A series of tests were ran on both the i7 920 and 875k at both stock speeds and at 4.0Ghz for comparison purposes. All test suites were left on stock settings.

3DMark Vantage
A standard in benchmarking. PhysX was left on as both processors were being tested with the same GTX 260 at the same speeds.

The 875k outscores the 920 at stock speeds. While they have the same total score at 4.0Ghz, the 875k has a higher CPU score.

Maxon Cinebench R11.5
The newest release of the popular Cinebench benchmarking suite which simulates what results you can expect using Maxon’s Cinema 4D suite. Great for multi-core CPU testing.

The 875k out performs the 920 at stock, and performs nearly as well at 4.0Ghz.

Fritz Chess
Testing how fast the CPU can calculate chess moves using a real chess game engine. Fully supports multi-core processors.

The 875k beats out the 920 at both stock speeds and 4.0Ghz.

Hyper Pi
Another favorite amongst overclockers and benchmarkers. Calculates how quickly the CPU can calculate Pi the 1m digits.

Although faster at stock speeds, the 875k is slower to calculate at 4.0Ghz.

POV Ray
Free CPU based ray tracing software. It has a built in benchmark for painless testing.

875k takes the cake in both cases.

ScienceMark
A cluster of real world, rather than synthetic, tests. Composite score is compared.

At stock speeds, the 875k scores significantly higher than the 920. At 4.0Ghz, the 920 marginally beats the 875k.

SiSoft Sandra
Although Sandra has a slew of synthetic tests, only a handful of the CPU tests were ran.

At stock, the 875k scores better in all tests, but it’s a toss up at 4.0Ghz.

Folding@Home
Real world folding results, running -smp 7 (A3 units) on the native windows client.

Although points vary by work unit, the A3 units all produce approximately the same points per day, give or take a few hundred. The 875k is the clear winner at stock speeds, but the 4.0Ghz crown is up for grabs.

Games
Games were ran at 1920×1080, and game preset high settings.

World in Conflict

Battlefield Bad Company 2

Gaming is a toss up, both the 920 and the 875k perform very well. 875K better in WIC, 920 better in BFBC2.

The performance of the i7-875k chip is impressive; it scores close to, or better than, the i7 920 in most tests and a games, but consumes much less power with a TDP of 95w and a max of 89 amps drawn. Since this in an unlocked chip, both the TDP max amperage can be changed in BIOS that support it. To me, that makes the 875k a clear winner.

Now for what everyone wants to see, overclocking these new i7-875k chips.

Here are two CPU-Z screen shots at stock speeds and default BIOS settings. These chips have a base clock of 133Mhz and multipliers of 9-22, 24 in turbo mode:

To show how easy it is to overclock this chip, let’s start with a video.

Okay, that was pretty good, but I know this chip can do better. Now, there are four different turbo multipliers, each corresponds to how many cores are being used. So if you want to hit different speeds or save power while using 1, 2, 3 or 4 cores you can. For my testing, I set all four multipliers the same to ensure I hit the speed I was looking for.

One thing to note, memory frequency is linked and synced with BClk, so if you want a higher RAM frequency, you will need to use a higher BClk, not a higher multiplier.

On to the overclocking fun.

First I decided I was going to keep the multipliers at the default 24 and just increase the base clock frequency. Increasing BClk to 180 yields a core speed of 4.325Ghz at 1.376 volts, not too bad on air!

How about just increasing the UNLOCKED multipliers? Well, I cheated a little on this one. I wanted to hit at least 4.3Ghz, but I was having trouble high multipliers, so I had to increase BClk by 2Mhz. Bumping up the multipliers to 32 and the BClk to 135 yields a clock speed of 4.324 @ 1.408volts.

My favorite method, increasing both BClk and unlocked multipliers. Setting the BClk at 162 and multipliers at 26 yields just over 4.2Ghz at 1.32 volts. With such low voltage, there is more overclocking headroom.

Considering these results were all on air, I am very impressed by this chip. It is, by far, the easiest processor I have overclocked. The unlocked multipliers really give it an edge over locked chips; having the option of mixing increased multipliers and base clocks provides and overall easier overclocking experience. Also note, overclocking varies from hardware to hardware, person to person. My experiences are not guaranteed reproducible.

Now that I have unlocked and overclocked, my mission has come to an end I can honestly say I love this processor. As a folder, I have been fond of the i7 920 for quite sometime now, but with the unlocked multipliers and power settings, I have found a new processor for my folding rig.

One thing that has not been mentioned yet, because it’s impossible to measure, is the overall smoothness of computing. For awhile, I have been noticing my 920 is a bit more jittery than I would like. A little lagging that is probably not that noticeable to most has really been bothering me, especially after switching between the 875k and the 920. The i7-875k is smooth as butter, I have not experienced any of the choppiness I have noticed with my 920.

All in all, the i7-875k is a keeper. I had been looking for a reason to move to the 1156 platform and now I have found it and couldn’t be happier.

Here’s your chance to get on the overclocking fun!

Intel has been kind enough to provide a similar set of hardware to give away to one lucky reader. It should be obvious what makes the i7-875k so special, but if you are unsure what it is, I will give you a hint. I have scrambled one word that describes the specialty of the i7-875k, you will have to unscramble it yourself, but it is easy! Here it is:

ueokcnld

Once you have it unscrambled, head over to the contest section of the forums for more details.

Disclaimer: Intel provided the i7-875k, DP55SB and kits of Patriot memory to be used in this review. They have also covered the cost of promotion of and prize for the contest in which one reader will win a similar package.

• Compatible with K-, J-, T- and E- thermocouples (others should work but these are the recommended types)
• User programmable offsets
• Internal memory stores 100 sets of temperature readings that can be transferred to PC software
• Data logging software for real time testing (good for working with fan and component placement and viewing their effects in real time)
• Ability to export data sets from software in .xls format to do comparative studies and graphing

The graph capabilities of the 72-7712 software are not phenomenal, it does however serve the purpose. Though dual software readout (T1 & T2) would be preferred; the logging capacity and decent feature set, as well as an Excel export feature make up for the software weaknesses.

Thermocouple 1 reading

Thermocouple 2 reading

Thermocouple 1 - thermocouple 2 reading. Temperature difference. This screen is most effective when trying to move case temperatures closer to ambient room temperature.

Example of exported data to .xls format

Thermal conductivity of the heat sink material is an important factor in air cooling. Copper and aluminum are the most widely used materials in PC HSF (heat sink & fan) construction. The properties of these two materials are critical to proper cooling of the processor.

The Chart below shows the thermal conductivity of materials for comparison. The only three that matter for this testing are aluminum, copper and air (water and the other items may be of interest to those who like to get a little wet).

:More information about thermal conductivity and conductive heat transfer:

Some simple ideas for improving the PC enthusiast experience:

Checking the case for hot spots.

Keeping your entire case as close to ambient is probably the most important thing that can be done to keep the HSF operating at its maximum efficiency. A heat sink can not lower temperatures below case ambient and will usually level out 4-12 degrees centigrade above case ambient no matter how much money is spent on it.

By identifying hot spots, proper fan placement can be made. Though these areas may not seem relevant to CPU cooling; they are. Air circulating throughout the case creates eddies, (a current of air running contrary to the main current; especially: a circular current : whirlpool) which in turn, remain hot and by cross circulation make  air circulating around them heat up.

Working in a similar fashion to the eddy, dead zones (hot area where there is no mechanical air circulation) may seem harmless, it is critical to circulate or eliminate this air to alleviate convection (heat transfer in a gas by the circulation of currents from one region to another). For dead zones a fan may not be an option and directed air may be needed. If directed air is not possible then closing in/sectioning off this area may be the only option.

Knowing where the hot areas of the case are allows for fixes that otherwise would not be possible.

Testing for efficiency.

Methodology: Air can only dissipate a fixed amount of heat due to its low thermal conductivity. Having a material of higher thermal conductivity does not always mean better temperatures, but it does allow a potential for lower temperatures, depending on other contributing factors. Testing the two most common heat sink materials to see these differences helps gain an understanding of what the conductivity numbers really mean.

Copper and Aluminum heat sinks tested for conductivity.

Copper; 56.8 seconds to reach maximum efficiency with a variance of 3.9 degrees centigrade

Aluminum 59.8 seconds to reach maximum efficiency with a variance of 7.9 degrees centigrade

This is the point where temperatures stabilize and heat is dispersed through natural convection. This is not a scientific test as the blocks were not exactly the same and some variables were omitted. What it does show is that copper will transfer heat faster and more evenly.

A double boiler is used to allow for better temperature control

Test equipment and stop watch used (phone) for testing. A Tenma 72-8540 is used as a control.

The copper and aluminum heat sinks used for the test

A two minute test of both materials (copper and aluminum) showed a 3.7 degree centigrade variance, copper being hotter (this is good, it means it will draw that much more heat to be dissipated). It must be taken into consideration that these heat sinks did not have a fan and the variance would have been lower during operation.

The results of this test correlate directly to the previous test results.

Testing your Heat Sink and Fan assembly

Using an Arctic Cooling AF64 PRO

If the HSF is not equalizing temperatures within a reasonable variance or running 10+ degrees above ambient case temperature (check the temperature at the intake area of the HSF to eliminate the possibility of a hot spot causing the problem) then a re-seat of the HSF may be needed and possibly a replacement HSF of higher quality may be in order.

Using information gathered with a good temperature meter will help guide the process of lowering case temperatures and in turn allow for a cooler processor, memory and hard disk drive.

Shots of the 72-7712

All display elements

Temperature readout screen

Variance screen

Setup: Offset adjustment screen

Front view of meter showing controls

Conclusion

Using a dual probe temperature meter with capabilities comparable to the 72-7712 is a definite step up from the volt meter type single probe units that were used in the past. With the data logging capabilities and other features available with this unit it is much easier to maximize case cooling and potentially gain a few hundred MHz from a heat limited overclock.

With acceptable quality, useful software and features the 72-7712 makes an excellent addition to the tool box of the overclocker or small PC mod shop.

]]> http://www.techreaction.net/2010/06/01/tenma-72-7712-dual-chanel-temp-probe-what-happens-with-tempratures-inside-a-pc/feed/ 4 http://www.techreaction.net/2010/05/06/amd-phenom-ii-six-core-1090t1095t-thuban-review/?utm_source=rss&utm_medium=rss&utm_campaign=amd-phenom-ii-six-core-1090t1095t-thuban-review http://www.techreaction.net/2010/05/06/amd-phenom-ii-six-core-1090t1095t-thuban-review/#comments Thu, 06 May 2010 10:23:51 +0000 thebanik http://www.techreaction.net/?p=6384 Introduction

Eight months ago AMD introduced their 2nd generation of Phenom 2 X4 Processors that was designed to use the AM3 socket while remaining backwards compatible with previous AM2/AM2+ motherboards. Although the Phenom II X4 965 was AMD’s flagship desktop processor, it wasn’t the performance that glued users on to it but rather the value for money it brought to the table. Alongwith processors like X 2 550/555 which could be unlocked to 4 cores depending upon the combination of chip and motherboard. Whether your allegiance lies with AMD or Intel, there is no denying the fact that competition is a good thing for us the consumers. Now you would ask, why should Intel affiniatos be happy about AMD’s hexa-core processor launch. Intel has a hexa-core Processor, the i7 980X aka gulftown but it retails for 1000$ and out of reach for majority of the enthusiast who are in their sane minds and without any roadmap for launching an affordable hexa-core CPU. We will find out in the next few pages whether the new hexa-core processor from AMD 1090T Black Edition pressurises the competition enough that Intel would be forced to do price cuts or maybe new product launches.

The new Phenom II X6 1090T Black Edition that we are benchmarking today operates at 3.2GHz with a 16x clock multiplier (200Mhz slower than the 965) but with a turbo speed of 3.6Ghz which should make non-overclockers very happy because when using single thread applications it means free performance. This new Phenom is based on the new new Thuban architecture and features 9MB L3 cache, with each core having its own dedicated 512KB of L2 cache (3MB total L2 cache).

And before we forget its not just X6 1090T which is getting launched today but there are other processors as well. Do find the basic specs of the processors listed below.

1090T Featureset

Lets have a quick look at the die picture of the 1090T. We wont waste your time with blabbering on the conclusions drawn from the picture.

*Die Picture

• True Six Core Processing

• AMD Turbo CORE Technology

• L1 Cache: 128KB (64KB Instruction + 64KB Data) x6(six-core)

• L2 Cache: 512KB x6(six-core)

• L3 Cache: 6MB Shared L3

• 45-nanometer SOI (silicon-on-insulator) technology

• HyperTransport™ 3.0 16-bit/16-bit link at up to 4000MT/s full duplex; or up to 16.0GB/s I/O bandwidth

• Up to 21GB/sec dual channel memory bandwidth

• Support for unregistered DIMMs up to PC2 8500 (DDR2-1066MHz) and PC3 10600

• Direct Connect Architecture

• AMD Balanced Smart Cache

• AMD Dedicated Multi-cache

• AMD Virtualization™ (AMD-V™)Technology

• AMD PowerNow™ 3.0 Technology

• AMD Dynamic Power Management

• Multi-Point Thermal Control

• AMD CoolCore™ Technology

In AMD’s words – “AMD Phenom™ II X6 Processors were designed for extreme megatasking, multi-threaded applications, and entertainment. To enable you to do more than you’ve ever imagined: create, edit, render, encode/decode and transfer dense HD content while watching HD content, burning CDs or DVDs while downloading music and video.” We certainly will see in the upcoming pages if they have been successful in their endeavor or not. But first lets look at the difference between the various new chipsets AMD has come out with.

New AMD Chipset comparison and architecture

And before we move on just so that you get to know a bit more about the 890FX chipset, do check the architecture of 890FX below.

System Specs and Benchmark List

Due to shortage of time, there is only a limited amount of setups that we have for comparison. But I guess this should be enough to see where does 1090T stands when compared to its existing product line and the target competition.

We would be comparing the setups with below mentioned benchmarks

Multithreaded 2d Benchmarks

Wprime

Everest CPU Queen

Winrar

Main Concept Encoding

PCMark Vantage

Memory Bandwidth Benchmark

Sisoft Sandra

3d Synthetic Benchmarks

3dMark06

3dMarkVantage

Cinebench

Games

Crysis Warhead

FarCry 2

Turbo CORE

Lets talk about the new feature in the x6 series of CPU, in my understanding Turbo CORE is something which plays with CPU’s TDP headroom when three or more cores are idle by automatically boosting the remaining three cores and throttling down the idle cores. On the Phenom II X6 1090T, 3.2 GHz CPU it can boost the CPU up to 400Mhz and put the final speed to 3.6Ghz. Quite impressive if you ask me.

Unfortunately it seems to me this is simply a marketing gimmick or fail engineering. I tried running a couple of single threaded programs like super-pi and pi-fast at stock speeds of 3.2Ghz but at no point of time any of the cores jumped to 3.6Ghz. At the max one of the cores jumped over to 3.4Ghz only to fall back on 3.2Ghz and at no point of time did any of the cores throttle down.

Though I should also add, that I was not using a 890FX board but I dont think Turbo Core is chipset restricted or atleast it simply should not be. On another thought it might be bios dependent and needs a motherboard bios update??? We will wait for the verdict to come out on this in the next couple of weeks.

Overclocking

This is the section which excites me the most, afterall I write reviews so that I can have access to the latest hardware(whoops!!! spilled the secret out). Lets then push it to the limits, ofcourse only on air. But there is more to come, LN2 results should be out in a week or two.

Before we begin let me tell you that the sample we have is a retail unit so though we would expect other processors to reach to the same level or better but many factors come into play when overclocking. First off we tried with the stock cooler which has seen some improvement from the 965’s stock cooler. Hold on to your breath, we were able to reach speeds of upto 4.4Ghz which was only 2d stable though. 4.2Ghz was 3d stable on stock cooler under an airconditioned room. Hugely impressed with the initial overclock on the stock cooler we slapped on a sample of CoolerMaster V10 cooler that we had. Though temperatures decreased but our overclock didnt improve. So next was disabling cores (remember this wont give much performance boost to most users, but can be done just to get maybe a second faster time in superpi). With the Gigabyte motherboard we had we were able to disable all but 2 of the cores from Bios. And with vCore pushed to 1.6V we were able to achieve 4.6Ghz which was stable enough to run superpi. Afterall things said and done, now the reviewer can go to sleep in peace.

Our retail review sample was able to reach 4.6Ghz with only 2 cores and 1.6V, but for 24/7 setup we were able to achieve 4.2Ghz at 1.55V. This is certainly good news since in our opinion 1090T would have a higher overclocking headroom than the 965BE processors out there judging by our retail review sample while adding 2 extra cores to it.

WPrime

We will be starting our tests with this multi-threaded benchmark application which gains from both higher frequency speeds and more cores as well. We will be running 1024m tests for this one.

The stock 1090T packs a great punch at stock clocks when compared to the Intel offering, however when both the processors are overclocked to 4.2Ghz, timings are almost neck to neck, with Intel i7 920 having a slight edge.

Winrar

Next up in the benchmark is the universal compression tool Winrar. We are not going ahead with its synthetic benchmark but we would be compressing a 500MB folder in each of the system and measure the time it takes to compress the folder. Lesser the time better the performance.

Winrar – Compression 500MB Folder

Looks like 920’s HT paying off here, Intel has a healthy lead here both at stock and when overclocked.

Sandra

SiSoftware Sandra is one of the most popular synthetic memory benchmark utilities. Sisoftware Sandra (System ANalyzer, Diagnostic and Reporting Assistant) is an information & diagnostic utility. It should provide most of the information (including undocumented) that you need to know about your hardware, software and other devices.

Here we see Read and write speeds of the memory which is set at 1600Mhz Cl8-8-8-24.

Intel with its triple channel memory has a huge lead here as expected both at read and write speeds. And 1090T sees gain over the 965. No surprises here.

EVEREST

Lavalys EVEREST is a system diagnostics and benchmarking utility that provides system and overclock information, advanced hardware monitoring and diagnostics capabilities, including tests for the memory subsystem. It also includes various CPU benchmarks, one of them is CPU Queen which is what we are going to use today. This particular test is sensitive to both frequency and active number of cores.

Everest CPU Queen

Another synthetitic benchmark, but here 1090T takes a huge lead over Intel i7 and the amount of lead is surprising for sure. If only in real life applications we could have seen such leads by 1090T.

PCMark Vantage

PCMark Vantage is a PC benchmark suite designed for Windows Vista and now Windows 7 offering one-click simplicity for casual users and detailed, professional grade testing for industry, press and enthusiasts.

A PCMark score is a measure of your computer’s performance across a variety of common tasks such as viewing and editing photos, video, music and other media, gaming, communications, productivity and security.

PCMark Vantage

Once again we see Intel scoring a huge lead over the hexa-cores even at stock clocks.

Main Concept

Main Concept is a multi-threaded video encoding application. We used a downloaded MPEG-2 file and converted it to H.264 format on all our systems.

Main Concept Encoding MPEG-2 to H.264

AMD 1090T outperforms i7 920 easily on stock clocks. But when matched clock to clock 920 takes a small lead. As usual we see much better performance from 1090T when compared to 965BE.

3dMARK06

Although this benchmark is a couple of years old, it still hasn’t lost its sheen. 3DMark06 is a PC benchmark suite designed to test the DirectX9 performance of your graphics card. A 3DMark score is an overall measure of your system’s 3D gaming capabilities, based on comprehensive real-time 3D graphics and processor tests. We ran only the CPU Test to measure the performance of our CPU’s.

3dMark06 – CPU Test

We already knew that 3dMark 06 favors Intel CPU a lot. Though at stock clocks 1090T soared ahead of i7 920 but again when compared clock to clock at 4.2Ghz, i7 920 took the lead back. 965BE as expected is at the bottom of the table.

3dMark Vantage

The latest installment (some say the big brother of 3DMark 06) of the famous 3D benchmark. This version added support for DX 10. This pushes the graphics sub system a lot harder than its younger brother and gives an idea of how your machine would perform under stressful games:

3dMarkVantage – CPU Test

Its Intel all along. i7 920 takes the lead even at stock clocks, which is almost 600 Mhz lower than the clocks of 1090T and further increases the lead when running at 4.2Ghz. 965BE retains its bottom most position.

Cinebench

CINEBENCH is a real-world cross platform test suite that evaluates your computer’s performance capabilities. CINEBENCH is based on MAXON’s award-winning animation software CINEMA 4D, which is used extensively by studios and production houses worldwide for 3D content creation. CINEBENCH is the perfect tool to compare CPU and graphics performance across various systems and platforms.

Cinebench

This is where 1090T really shines. At stock clocks 1090T has a healthy lead. Though when compared at 4.2Ghz, i7 920 reduces the lead quite a bit but still the leader is 1090T.

Crysis Warhead

Crysis Warhead

We see a marginal improvement after overclocking the CPU. But not much difference to pick between Intel and AMD. Pick any and you will be satisfied as far as your gaming needs are concerned.

FarCry 2

Far Cry 2

Pretty clear that Far Cry 2 is not CPU limited but GPU limited atleast in our case. If gaming is your concern then you can pick any one of them and you will certainly not be disappointed.

Price and Conclusion

1090T and 1055T are the first real hexa-cores which is in reach of 99% enthusiasts, since Intel’s 980X retails for around 1000$(more than Rs. 55000 in Indian retail market), though if you are looking at purely performance numbers then I am sure 980X has simply no competition. We will have a 980X as well for comparison very soon, so stay glued to Erodov, :p. As of now 1090T is selling for around Rs. 15000 and 1055T for Rs. 10100 from SMC International and other retailers in India.

Though 1090T is impressive but which group of enthusiast should really be looking forward to upgrade?. I will say anyone involved in video work, photoshop, rendering would be my first target group. AMD’s new flagship processor is easily able to keep up with the Intel’s 920 and in some test better it. Gamers I would say would not really see much benefit from either going to Intel or AMD’s hexacore just now. There is really no benefit of having extra cores when they are not being used for gaming. If you are a gamer and want more performance out of your C2D or Phenom 2 computer then will suggest a SSD as an upgrade for sure, specially if you play multiplayer games. Ofcourse if you are a overclocker at heart and would like to buy a future proof system for the next 2-3 years then 1090T has plenty to offer. Our system was able to do 4.6Ghz for a very small fraction of time for some 2d benchmarking, 4.4Ghz for benchmarking all cores with 2d benchmarking softwares like wprime, and 4.2Ghz for all our benchmarking suite. It will be interesting to see how much it can do under Liquid Nitrogen which we will be doing most probably this weekend. We have been told that these hexacore’s better overclocking headroom is mostly because of GlobalFoundrie’s addition of low-k dielectric in their 45nm manufacturing process, because of which these chips leaks less current, drawing less power and outputting less heat. Just uptill a few months ago, AMD had a 140W quad core and today they are delivering an hexacore within a 125W power envelope. Thats quite a feat if you ask me.

In the end, we are at a crossroad where 1090T has brought AMD much closer to Intel than they have ever been in recent times. Its pretty hard to clearly announce a winner and to recommend one company over the other but whichever company you choose to build your rendering/gaming/enthusiast system, you’re probably not going to regret it.

We would like to thank Mr. Saini from SMC International and AMD for providing us with the review sample.

Also be sure to check out TestFreaks for more reviews on the AMD Phenom II X6 1090T or Intel Core i7 920.

That’s right! FREE HARDWARE! Just for being a registered and active blogger of TechREACTION.net.

How do I sign up? Simple, first register at this page to become a TR blogger. Once you receive your account information, half the job is done.

Second, go to the TechREACTION forums and register there as well. You must be registered in both places to be eligible.

Bloggers are given special access to a private forum where they can discuss content and blogs, as well as be notified of available review samples and other perks.

What about the giveaway? It’s easy. All you have to do is be yourself; a technology enthusiast. To qualify you must follow these simple rules:

- One blog post equals one submission to the contest.
- Two blog posts equals two submissions, and so on, and so forth.
- In cases where there are multiple prizes, a single blogger will only be able to win one. Prizes are drawn at random.

I have never blogged before, what’s the deal? Don’t think of it as something only for the elite. Blogging is really no different from posting on a forum, such as this. The difference being that it’s more organized, looks more professional to the reader, and puts the information on the front page rather than just buried deep in some thread somewhere.

Many average Joe’s who simply starting out with some basic blogging are now writing doing full fledged reviews and making a solid name for themselves, and you can too.

I’m interested, but what can I blog about? Everything, as long as it’s technology related. Here are some examples:

Just did some benching with a few buddies or by yourself? Blog it.
Bought a new smart phone or mobile device that you can review? Blog it.
Testing out the new version of an operating system? Blog it.
Seeing what better cooling does for your CPU overclock? Blog it.

The list goes on and virtually anything can be blogged as long as it’s related to the main subject: Technology.

Giveaways occur once a month, so by registered for this months giveaway you are in fact registering for every giveaway from here on to infinity. Sounds easy enough no?

Sponsored By AMD

Previous Contest Prizes & Winners:

April Prize: Thermalright Heatsink Pack (TRUE & 2xTRAD2 GTX’s)
Winning Blogger: Burn

March Prize: EVGA P55 Motherboard
Winning Blogger: carpo93

February Prize: Crucial Ballistix Tracer DDR3 Kit & OCZ Gladiator CPU Cooler
Winning Bloggers: 0Ro!, The Duke

December Prize: Corsair HX750W & Razer Lachesis Gaming Mouse
Winning Bloggers: The Duke, Archer

November Prize: Gigabyte P55-UD4P Core i5/i7 Motherboard
Winning Blogger: thebanik

October Prize: 2x AMD Phenom II X4 965 Black Edition
Winning Bloggers: mav2000, Neuromancer

Congratulations to you!

The 555, like its predecessor the 550, is a dual core AMD processor based on the Phenom II architecture.  The original 550BE was released with a half step lower multiplier and was based on the C2 stepping, although a subsequent C3 stepping has recently appeared on Newegg as a tray processor also a half step reduced for a 3100 MHz Core frequency.  Both are Black Edition processors, meaning they are multiplier unlocked upwards, and if the 550 is an example of 555 performance, we can expect great things from this chip.

The 555 Black Edition is a C3 stepping core reduced 955/965 edition quad core processor.  With the right motherboard and a little luck it is possible to unlock them to quads (4 core processors) or trikes (3 core processors). It is not a given though, and at the price point (~$100) if you want a quad core, the 955 C3 stepping quad is $160.  Buy it and be happy.  Or buy this processor, if all you need is a high speed dual core, pat yourself on the back if it unlocks, but do not complain if it does not.

As an overclocking enthusiast, and a proclaimed Hardware Junkie, I usually know what to expect with a given architecture before it is ever in my hands.  As a budget minded consumer, I also know that more is not always better.  So $100 for a dual core processor from AMD is a lot of money.  Especially with the L3 cache-less Athlon II x4 quad core processors selling for the same price.  Benchmarks show that there is not a lot of difference between the two architectures so why bother?

Archer of techreaction.net and overclockers.com did a great review of AthlonII vs PhenomII architecture.  As you can see in the tests that he ran, it was fairly even until he put multitasking on the table. As a subjective reference, I recently had the opportunity to run a few athlon II chips myself, and they feel ‘slow’. I will start another blog explaining this as this thought process is along and arduous one that carries over 10 years of enthusiast PC usage.

Back to the 555.

Series Phenom II X2

Model HDZ555WFK2DGM
CPU Socket Type Socket AM3
Tech Spec, Core Callisto
Multi-Core Dual-Core
Name Phenom II X2 555BE
Operating Frequency 3.2GHz
Hyper Transports 4000MHz
L2 Cache 2 x 512KB
L3 Cache 6MB
Manufacturing Tech 45 nm
64 bit Support Yes
Hyper-Transport Support Yes
Virtualization Technology Support Yes
Voltage 0.875V – 1.40V
Thermal Design Power 80W

The chip was used on two different motherboards, the high end Crosshair III formula under water cooling in winter.  Again on the gigabyte 890GX under Dry ice. This was my first attempt and dry ice and I had not yet learned the intricacies of extreme cooling.  It was a borrowed aluminum pot with a too small core to meet the size of an AMD IHS.  Temps swung 40C under load, and this was partially due to using alcohol instead of acetone.  I will be retrying later under DIce with an F1 pot and acetone as the sublimating medium.

The water run was poor, and AOD was not usable on the CHIII for some reason so I could not use the normal validating trick of dropping multi and bringing it back up to pass validation

Under a poor dry ice run at only –45C I was able to pull this off

4628.5 CPUZ validation link

5122.83 CPUZ validation link

Not the best of group but I will be working on it soon.  4.6GHz was definitely good for water  cooling I expect to see a bunch of 4.2-4.4GHz air cooler soon despite the coming warmer months in the US.

My initial impressions are that this chip is one monster overclocker, and I am apt to call it, an E8400 in sheep’s clothing.  I will post back when I reach 6GHz.

My goal was to better understand the impact of adding a second 360 radiator, which in theory doubles the heat dissipation capability of your loop. But what does it bring to the table in terms of overclockability on the AM3 platform?
Read on for a lightweight comparison between running one and two 360-sized radiators in a single loop, cooling a quad core AMD Phenom II CPU………….

TEST SETUP

Liquid Cooling components

Radiator(s): 1-2x Alphacool Nexxxos Xtreme III rev. 2
CPU block: EK-Supreme Acetal
Pump: Swiftech MCP655-B
Reservoir: Magicool 400mm POM tube
Tubing: Masterkleer 12,7/15,9mm
Fittings: 13/16mm compression
Radiator fans: 120x120×25mm push/pull 7.9v
Coolant: Pure distilled water
Loop order: Pump -> Radiator(s) -> CPU -> Reservoir

PC hardware & software

Motherboard: Gigabyte GA-MA790FXT-UD5P (BIOS f3m)
CPU: AMD Phenom II x4 965 BE (rev. C3)
RAM: 2×2GB OCZ Platinum (1600MHz 7-7-7 1.9v)
GPU: 2x XFX ATI HD 5770 1Gb (rev. 1)
PSU: Chieftec CFT 1200G-DF 1200W
Software: Windows 7 Ultimate 32bit
CPU-Z v1.53
Prime95 v25.9
HCI MemTest v4.0
Core Temp v0.99.5

MAX stable OC
single radiator

Below are a CPU-Z screenshot of the highest achievable OC using a single 3×120 radiator. Prime95 blend test was used for load and stability testing.

4116MHz CPU (1,55v) & 2940MHz CPU-NB (1,50v)

I finally settled for a CPU frequency of 4116MHz, together with a CPU-NB running at 2940MHz. With a load temperature of 51C°, there is no headroom left to push things further.
And as expected the dual radiator setup had no problems at all with this heat load.

MAX stable OC
dual radiator

CPU-Z screenshot of highest achievable OC using dual 3×120 radiators connected in serial. Stability where once again confirmed with the help of Prime95.

4158MHz CPU (1,55v) & 2970MHz CPU-NB (1,50v)

With two radiators connected in serial, it was possible to push the frequencies even further. Managed to squeeze out 4158MHZ for the CPU and 2970MHz on the CPU-NB.
With temperatures shooting over 52C° before crashing, the single radiator setup just couldn’t to handle this kind of heat load.

QUICK FINAL THOUGHTS

Moving from a single to dual radiator setup allowed for a small but respectable increase in frequency. To be precise, an extra 42MHz on the CPU and 30MHz on the CPU-NB. Not too shabby at all in my opinion! ….

]]> http://www.techreaction.net/2010/03/11/single-vs-dual-3x120-radiator/feed/ 3 http://www.techreaction.net/2010/02/18/temps-watts-cpus-and-extreme-cooling-part-1/?utm_source=rss&utm_medium=rss&utm_campaign=temps-watts-cpus-and-extreme-cooling-part-1 http://www.techreaction.net/2010/02/18/temps-watts-cpus-and-extreme-cooling-part-1/#comments Fri, 19 Feb 2010 04:57:23 +0000 Buckeye http://www.techreaction.net/?p=4448 In the beginning when I first started using different methods of extreme cooling the first unit I tested was a Jinu Single Stage Cooling unit that was custom made for me. This was an amazing cooling unit that included a Single Stage Phase Chiller installed in a TJ07 case.

The TJ07 cooling tower sat next to what I called the main rig which was in another TJ07 case. It had a EVGA 790i SLI motherboard and a QX9770 which ran at 4.8ghz overclock for 24/7 use. It was an amazing machine that had so much power at the click of the mouse button to run anything I wished at an instant.

Later as I became even more interested in extreme cooling and searched for even better ways to cool things. This amazing machine enter my work area. Built in 2006 by Jinu, this 2 Stage Cascade was the most powerful cooling unit I had, and I love every minute of using it.

Over time with using these machines you find that they have pro’s and con’s. On the good side, they get very cold and can handle high watt loads to keep overclocks stable. On the bad side they can generate lots of heat and can make a fair amount of noise like the 2 Stage Cascade.

The people who build these units, for the most part,  understand how they work, at least some do, but often designs are copied that work with out much regard to how and why things done.

Designs come and go, some build different units called benching units while others are smaller and built to fit in a case with the computer.

The problem that every unit must address is the heat coming from the CPU while its running in a overclocked state. Load testers are constructed that can be set to a given watt load such as 300watts and the unit is tuned for that heat load.

Tuning can be a tricky business that involves cutting cap tubes, charging the unit with refrigerant of a given type to a whole host of other ways. Different parts such as the Evaporator can play a large role in the capabilities of the unit and how it handles a heat load.

Here are three different types of Evaporators. A Chilly1 on the far left, a Teyber Stepper in the middle, and a Jinu Maze type on the far right.

There are a whole host of different types of Evaps and they vary in design from builder to builder. But which type works the best for a given cooling unit ?

So far that I have seen, there is no way to actually measure the effects of these things besides throwing a load tester on the unit and set to a watt load, or by what a builder says works well or doesn’t, with no real way to back that statement up with data other than experience of the builder and what they say.

Some of the things I am interested in to study are effects of Evaporators on units, how fade can be addressed and what does the heat output of a CPU really look like during a test.

I began building a setup that could monitor both temps and watt loads at the same time and show those readings in graph form so you could see just what was happening during a test.

Now there is problems when attempting this. One of the problems is how can I measure the heat output of just the CPU while the computer is running ?

Besides using a load tester, measuring the temps at a given watt load like what was done for me when I picked up the Jinu 2 Stage Cascade.

At 200w… it is holding -95c…
At 270w… it is holding -89c….
At 285w… it is holding -87c….

So that is a start and gives me an idea on what to expect.

I have several phases to this project that I am working on. One is I need to build a load tester and I have some parts for that here now and awaiting the rest.

I will use the load tester to calibrate my findings, or at least have some set data points that I can compare data measured with.

To monitor watt loads while the computer is running I will have to use something like a Kill-O-Watt meter like this one pictured and a unit that has a USB interface so the data can be feed to a computer and graphed with a Watts Up Pro unit.

To monitor temps I will use a temp probe that has a USB connection so data can be uploaded to the computer and graphed along with power usage.

So now armed with the ability to measure and collect data, send that to a computer and graph the incoming data I will begin my first tests.

The computer for the first part of this will be as follows:

Gigabyte GA-790FXTA-UD6, AMD 965 C3 CPU, ATI 4890 for graphic card and Corsair Dominator GT 2000mhz CL8 memory all powered by a Coolermaster Ultimate 1100watt PSU. For cooling I will use a Jinu Single Stage Phase unit.

The first test here shows the temperature drop of the cooling unit unloaded, that is it is not connected to anything.

I must note here that the units I am using for monitoring temps has a little lag to it from the USB/Computer interface, so it is not 100% perfect in the time scale and it does have an error margin in the actual temp readings. I watched the actual temps with a Fluke meter and these are with in +- a few degrees C.

So now I have a base line of how the unit handles without a load. When the load tester is finished I will add to this actual watt loads and what the temp reads at to make a performance curve.

Lets add the watt usage of the computer now.

The second test will show the start up of this setup at stock settings, the power usage of the full computer and how watt load and temps change from power on to sitting on the desktop.

First the Jinu Single Stage Phase unit must be turned on and temps pulled down to operating temperatures, then the computer will be started and run until it is at the desktop and idling.

CPUz screen of what the CPU is set at and its vcore setting.

Now the system is started up and run until its idling on the desktop.

So here you can see that the start up of the computer and how many watts it is putting out, it maxed out at ~280 watts during start up and then settled down to ~210 watts while running on the desktop.

Temps started at room temperature and the Phase unit dropped to ~-48c when the computer was started. You can see that temps warmed up a few degrees C then settled down to ~-48c for normal desktop operation.

The Phase unit showed it can handle these watt loads very well.

So now we have the basic parts of the up and coming tests. We can see in graph form what the power in watts is being used by the whole computer and how the Phase unit responds to the watt loads generated.

In the next series of blogs I will show bench programs run, how everything responds to these tests and how things react to higher over clocks.

A special thanks to Archer for the inspiration in doing this project

If you do a search on Newegg you come up with 36 listings which range in price from $145.99 to $283.99 with kits  in 6-7-6-18 to 9-9-9-27 timings, and 1.5v to 1.65v. So as you can see, this range of memory has a pretty wide selection of kits for what ever you are looking for.

The G.Skill F3-12800CL7T-6GBPI kit spec’s as follows: DDR3-1600 CL7-8-7-24 at 1.5v, PC3-12800 2GBx3. G.Skill Part number is: 10051040027518. So this kit will give you 3 sticks of 2gb each for a total of 6gb.

http://www.gskill.com/products.php?index=252

You can see a list of Qualified Motherboards there. Interesting to see that there are no EVGA Motherboards listed, but I am sure that will change.

Add a few pictures of the kit and my setup which is a Gigabyte GA-X58A-UD3R motherboard and Intel Core I7 975EE CPU.

The PI refers to the heat spreader which for these is black, they look very nice and are not to tall to interfere with most heat sinks. They are actually about a 1/16″ taller than some of the other kits I have so its not that much. The dark blue logos look very nice on a Gigabyte board for those who want colors to match up with there rigs.

Because there are a lot of kits to pick from in this memory range I wanted to keep my tests at manufacturer spec’s for the first round and then see what I could get them to do with some RAM overclocking.

So I setup the Gigabyte GA-X58A-UD3R & 975EE at stock defaults and set the RAM to 1600 7-8-7-24, 1.5vdimm running on SS Phase cooling and away I went.

First screen shows these results.

SuperPi 1M – 11.766s, PiFast – 24.61, wPrime 32M – 7.375s

Memory Read: 17231 MB/s, Write: 14066 MB/s, Copy: 19550 MB/s, Latency: 40.4ns

Then I up the overclock to 4.15ghz

SuperPi 1M – 9.906s, PiFast – 30.56, wPrime 32M – 6.063s

Memory Read: 17002 MB/s, Write: 14463 MB/s, Copy: 20970 MB/s, Latency: 39.0ns

And now 4.32ghz

SuperPi 1M – 9.453s, PiFast – 19.73, wPrime 32M – 5.89s

Memory Read: 16932 MB/s, Write: 14376 MB/s, Copy: 20671 MB/s, Latency: 39.1ns

And I topped out at 4.4ghz

SuperPi 1M – 9.25s, PiFast – 19.33, wPrime 32M – 5.687s

Memory Read: 17814 MB/s, Write: 14650 MB/s, Copy: 19522 MB/s, Latency: 38.6ns

So I was able to go all the way to 4.4ghz at 1600 7-8-7-24 and at 1.5v

Now overclocking this RAM was a bit of a problem. In order to reach a speed of 1800 I had to loosen the timings to 9-9-9 which resulted in poor stability so I left everything at 1600 7-8-7 which is where this RAM seemed to like to run.

SuperPi 1M – 11.703s, PiFast – 24.34, wPrime 32M – 7.235s

Memory Read: 19000 MB/s, Write: 15569 MB/s, Copy: 21878 MB/s, Latency: 39.ons

I spent just about a full day with Memtest trying to find sweet spots but going lower in timings would usually result in noboots or errors in Memtest. Again the sweet spot for this RAM is at manufactures spec’s, imagine that

When the first sets of RAM came out for I7 CPUs a good set of 1600 9-9-9 1.65v was plenty good for any rig. Now with manufactures that are able to get better results with kits for things like timing, speed and lower volts we are getting faster, tighter RAM that runs on less voltage, this is all a good thing imo.

But it remains to be seen that tighter RAM really helps any gaming platform, office apps or general use that anyone would really notice in everyday use. But for benching platforms and high end gaming rigs you would want higher end kits anyway.

This kit fits in well with a good mid range computer for everyday use. I am a big fan of lowering volts required to run tho and that will result in less heat output, plus help to save on your power usage.

So with a nice mid priced set of RAM this kit should be good to go on any mid ranged computer, but don’t expect much RAM overclocking with it. To me if you want to run RAM faster then get a set that is rated for that. High end gaming rigs would want to use the higher speed kits, why would you put anything less in a rig you spent a lot of money on.

At a speed of 4.4ghz that I was able to achieve before I had to start loosening up timings and giving it more volts I think that fits in well with high end air cooling and good water cooling setups, just don’t expect to overclock this RAM, use it at stock and you should be very happy.

You can find this memory at Newegg.

Thank you G.Skill for the sample.

Parts used in the test:

MSI P55-GD80 BIOS v1.7

2×2GiB Kingston HyperX 1600 CL8

Intel Core i5 661

Sapphire X1600Pro for CPU tests, MSI N260GTX Lightning for 3D Mark 03 test

Akasa 1200W 80+

Started to test the max BCLK.
After some restarts and VTT voltage rising to 1,392V I managed to get into Windows on 250MHz BCLK with 2 cores and HT enabled. Then I used the buttons for BCLK rising on the board and I got the max. validated BCLK of 253MHz.

After I found the BCLK limit I started to clock the CPU for max. frequency validation. First try I stopped at 5300MHz 1.64V. Damn!!! More no go. After a hour of freezing, rising voltages, lovering voltages a idea lighted up…what if the problem is the memory divider? And I was right! It has problem with 4x memory divider to run the memory more then 1750MHz…With divider 3x no problem. So a little afraid off the divider problem I went back to the max. frequency…and after some button pushes on the mobo I get a frequency of 5761MHz 1.64V, VTT 1.366V, PLL 1,55V.

validation link: http://valid.canardpc.com/show_oc.php?id=1001302

Then I played a little with SPI 1M, SPI 32M and PiFast. Here are the results:

SPI 1M: 7.656s – CPU: 5536MHz, MEM: 664MHz CAS6-7-6-21 1T

SPI 32M: 7m 39s 062ms – CPU: 5511MHz, MEM: 661MHz CAS5-7-6-21 1T

Hexus PiFast: 16.86s – CPU:5561MHz, MEM: 667MHz CAS5-7-6-21 1T

And at the end I run 3D Mark 2003 with MSI N260GTX Lightning clocked at 792/1584/1050MHz(Core/Shader/Memory) with the result:

68981 points – CPU: 5461MHz, MEM: 655MHz CAS6-7-6-21 1T

Thats all for now from single stage cooling… with the teammate we plan to run it on LN2 in a short time… so stay tunned