TechREACTION » wez

[Blog] Asus 5870 + XFX 5850 CrossFireX Mismatch

wez — Wed, 30 Jun 2010 18:32:40 +0000

ATI 5870+5850 CrossFireX

I’ve been playing around with two mismatched GPU’s in CrossFireX. Namely a reference model HD5870 from Asus and a custom PCB HD5850 from XFX. Perhaps not the most common setup, as most would much rather go for two identical cards. But if you happen to stumble upon the odd card, then why not give it a try, right?

Hardware & Software

Gigabyte GA-MA790FXT-UD5P
AMD Phenom II x6 1090T BE
Kingston HyperX LoVo 1866 MHz 2×2Gb
Asus ATI Radeon HD5870 1Gb and XFX ATI Radeon HD5850 1Gb
Chieftec 1200G-DF

Windows 7 x64 Ultimate
Catalyst 10.6 + 10.6 CFX profiles

System settings remained the same for all runs.

Asus HD5870 to the left and XFX HD5850 to the right.

Below you’ll find a few game runs together with 3DM06 and Heaven Benchmark. For the game benchmarks I used Fraps to record 30 seconds worth of data (MIN, MAX and AVG FPS) while navigating a set route in-game. The average values where then calculated from multiple runs, which is shown in the charts. I included the results from a single HD5870 as well, for reference.

Battlefield Bad Company 2
Single player level Upriver
1920×1200 / DX11 / Max quality / 4x MSAA / 16x AF / HBAO

Half Life 2
Single player level Water Hazard
1920×1200 / DX9 / Max quality / 8x MSAA / 16x AF

Left 4 Dead 2
Single player campaign The Passing, chapter 3: Port
1920×1200 / DX9 / Max quality / 8x MSAA / 16x AF

S.T.A.L.K.E.R. Call of Pripyat
Single player campaign, indoor and outdoor environments in Pripyat.
1920×1200 / DX11 / Max quality / 4x MSAA / 16x AF

3D Mark 06
Default settings
ATI HD5870

ATI HD5870 + HD5850 CrossFireX

Heaven Benchmark 2.0
1920×1200 / DX11 / High Shaders / Normal Tessellation / 4x AF
ATI HD5870

ATI HD5870 + HD5850 CrossFireX

Final Thoughts?

Visible gains in all tested games and benchmarks, so no doubt its a viable option running mixed cards. But would have been interesting to see how this setup would stack up against two 5850’s. I’d still recommend running matched cards where possible. But as mentioned at the top, if you happen to have an extra card just lying there collecting dust or get the chance to get one real cheap, then go for it!

[Blog] Kingston HyperX DDR3 ‘LoVo’ OC Testing

wez — Mon, 31 May 2010 13:47:50 +0000

Kingston HyperX DDR3 ‘LoVo’ OC testing

Just a quick look at a memory kit from Kingston, namely a set from their somewhat new ‘LoVo’ series. Judging by the name its a low voltage kit, so not a kit directly aimed at the enthusiast market. But word has it they might be worth a look after all!

According to their site, Kingston currently offers three versions of their ‘LoVo’ kits, 1333, 1600 and 1866 MHz. But there also seem to be an 1800 MHz kit listed at some shops, possible a replacement for the 1866 model? Either way, all share the same timings of 9-9-9-27 with recommended voltage ranging from 1.25 to 1.35v. And currently only available in 2×2 Gb dual channel kits.

The 1866 MHz kit is what I’ll mess around with today, part number is KHX1866C9D3LK2/4GX
This kit comes with two preset XMP profiles:

- Low Voltage 1866 MHz 9-9-9-27 @ 1.35v
- Ultra-Low Voltage 1600 MHz 9-9-9-27 @ 1.25v

Although for some reason the 1600 MHz profile didn’t get recognized properly. Now this could very well be a software issue on my part, but thought I’d at least give it a mention.

The sticks come with Kingston’s standard HyperX heat spreader, commonly used in a lot of their kits. But one thing is new, and that is the color!

Hardware & Software

- Gigabyte GA-MA790FXT-UD5P (BIOS f8h)
- AMD Phenom II x6 1090T
- Kingston HyperX LoVo 1866 MHz 2×2Gb (KHX1866C9D3LK2/4GX)
- ATI Radeon HD5870 1Gb
- Chieftec 1200G-DF

- Windows 7 Ultimate x64
- Prime95, Memtest, SuperPi, CPUz and Easytune

Not the most optimal platform to test high speed memory on. And low voltage testing is also out of the question as 1.6v is the lowest this board allows me to set. But ignoring that, let’s see what they can do!

These settings will all run Prime95 and Memtest for a few hours without any hiccup, so I’d call them stable.

Starting at 1.60v, 1624MHz

1.65v, 1688MHz

1.70v, 1712MHz

1.75v, 1720MHz

Voltage scaling beyond ~1.70v is pretty limited, although small gains are still there. Also, relaxing timings didn’t allow for much higher clocks, so I left those screens out on purpose.
Either way I think it’s safe to say that these sticks are capable of much higher clocks, but simply put the motherboard is holding them back. Hopefully I get a chance to test them out on a 890FX board in the near future, and I’ll be sure to post an follow up when I do.

And just for the sake of it, this is the limit for passing a SuperPI 32M run.

[Blog] 5770 CF vs. 5870 – Game Performance Comparison

wez — Sat, 20 Mar 2010 05:00:23 +0000

A pair of ATI 5770 1Gb in Crossfire vs. a single ATI 5870 1Gb. One could call it a fair fight considering the price point of both setups. They will be put through their paces in some of today’s game titles…

Test Setup

Hardware

Motherboard: Gigabyte GA-MA790FXT-UD5P
CPU: AMD Phenom II X4 965 BE (4,2GHz CORE/2,9GHz IMC)
RAM: 2x 2GB OCZ Platinum (1550MHz 6-6-6)
Monitor: LG W2600HP-BF 1920×1200 pixels
GPU: Two ATI HD 5770 1Gb (Crossfire) & one ATI HD 5870 1Gb

Software

MS Windows 7 Ultimate 32bit
Catalyst 10.3 February preview (driver package 8.71.3-100205a-095620E)
Fraps
Battlefield Bad Company 2
Aliens vs. Predator
Borderlands
Half Life 2
Call of Duty Modern Warfare 2

The FPS where recorded over a 20 second period using Fraps. Multiple runs where completed for each game, and the average values where then calculated for each game and GPU setup. Vertical Sync where disabled for all tests. And the highest possible quality settings where used, unless else is noted.

Battlefield – Bad Company 2
1920×1200, 8x MSAA, 16x AF, HBAO, DX11