Quad Core vs Dual Core: Q9550S vs. E8600, Part III – CPU Scaling with CrossFire
Test Configuration
Test Configuration – Hardware
*Intel Core 2 Quad Q9550S (engineering sample reference 2.83 GHz; overclocked to 3.40 Ghz, and 4.0 Ghz – i.e. 471 FSB: Please note: only for Clear Sky, FarCry2 and TC:X3, we used 469 FSB with CrossFireX-3 because of thermal/stability issues.
*Intel Core 2 Duo E8600 (reference 3.33 GHz; overclocked to 4.0 Ghz and 4.25 Ghz ).
* ASUS Rampage Formula (Intel X48 chipset, latest BIOS. PCIe 2.0 specification; CrossFire 16x+16x).
* 4 GB DDR2-PC8500 RAM (2×2 GB, dual-channel at PC6400 speeds or as close as possible using CPU:RAM dividers).
* ATi Radeon 4870-X2 (2GB, reference clocks) byVisionTek
* ATi Radeon 4870 (1GB, reference clocks) by ASUS
* Onboard SupremeFX-II (ASUS P5e Deluxe motherboard daughter-card)
* 250 GB Seagate Barracuda 7200.10 hard drive
Test Configuration – Software
* ATi Catalyst 9-2, highest quality mip-mapping set in the driver; Catalyst AI set to “Standard”
* Windows Vista 32-bit SP1; very latest updates
* DirectX November 2008.
* All games patched to their latest versions.
Test Configuration – Settings
* vsync is off in the control panel to “application decide” and is never set in-game.
* 4xAA enabled in all games and “forced” in Catalyst Control Center for Crysis and UT3; all in-game settings at “maximum” or “ultra” with 16xAF always applied
* All results show average, minimum and maximum frame rates except as noted.
* Highest quality sound (stereo) used in all games.
* Vista32, all DX10 titles were run under DX10 render paths
We found with later testing that our last tested Catalyst 8.12hotfix drivers and 9-1 are quite similar so in Part 3 we start with Catalyst 9-2 and with a much larger sample of actual game benchmarks than we presented in Part 2.
Great article. Here’s what I found most interesting
If we just look at the minimum framerates for the chips are similar clockspeeds (Q9550s @ 3.4 vs E8600 @ 3.33 and Q9550s @ 4.0 vs E8600 @ 4.0) the quad core comes out on top the majority of the time.
For the two-way Xfire tests, the Q9550 (at similar clockspeed to) beats the E8600’s minimum framerates in COD4, UT3, Lost Planet, HL2: LC, FEAR, ET: QW, WiC, FC2, and PT Boats. The two chips, for the most part, tie in the games Stalker, Crysis, and X3. The only game where the quad loses is Call of Juarez.
When we look at the three-way XFire tests, the results are basically the same except Lost Planet and PT Boats moves from the “win” category to the “tie” category for the quad core.
I wonder what’s up with the Call of Juarez results. Even with the chips at the same clock speed, the quad core loses fairly significantly. At 4.00 GHz, the quad’s minimum framerate is 31 while the dual’s is 42.
I wondered about CoJ as i was testing and repeated those benchmarks many, many times; far more than with any other of my tests. I would say that some of it is probably partly because of the Cat 9-2 drivers. If you look back on this benchmark to our September testing with Cat 8-1 all the way through Cat 8-12hotfix, there is definitely some variance with multi-GPU performance.
So let me theorize that there appears to be a ‘hitch’ in CoJ – you can actually watch it “stutter” in a couple of places – that the slower clocked Quad simply cannot overcome that appear to really skew the bottom [and thus average and max] framerates. It exaggerates what happens when you actually play CoJ, similar to my old STALKER benches that had way too high of a maximum as they panned the sky. The CoJ benchmark was also never updated, although the game was. That makes it somewhat flawed in my opinion, as the vendors are continuing to optimize for the game, not for the old benchmark. In the future, it will not be so important – as for example, in my current benching, “Vista 64 vs. Vista32-bit”, my Q9550s is at 4.0Ghz where this is not observed quite so much.
It also means that I am considering making a custom timedemo from the latest patched CoJ. I wish Techland would update theirs. Or maybe I will wait for “CoJ 2, Bound in Blood” and use that new benchmark instead. I am looking forward to its release, soon.
http://www.nugadgets.com/products/ProductDetails/68514Call_of_Juarez_2_PC.1496901.1.html
they say 1-3 weeks, but that is not official. The trailer says, “Summer”.
Here is a trailer on You.Tube:
http://www.youtube.com/watch?v=-CZi_FKsyPE
You also need to realize that CrossFireX-3 is still imperfect; you can see it’s scaling is still not “bang-for-buck”. Clearly there has been drastic improvements overall in the CFX-3 Catalyst drivers over the last 6 months, but there is plenty of room for more.
Yep, I can’t wait to see how multi-core CPUs and GPUs take off this year. Check out the following results for the new Tom Clancy game:
http://www.pcgameshardware.com/aid,679029/Tom-Clancys-HAWX-Benchmark-review-with-15-CPUs/Practice/
Those were some of the most striking results I’ve come across yet – even more striking than GTA4.
Hey you I’m a big fan of your blog. Hope you keep updating it regularly.