Tiger Woods Out...Until August!
 |
| |
| 19 User Comment(s) • 9 root comment(s) |
 Trogdor (39)  Oct 23, 2006 - 01:51 pm
| » Flawed Comparison
Sorry Brandon, but I think this is a flawed comparison. It amounts to comparing two different chipsets and motherboards. I'd much rather see the tests done on the same motherboard, the X16 capable board in this case but force the PCIe slots to X8 bandwidth. It would also be interesting to see how ATI cards stack up, although that would entail using AM2 since there are no dual X16 Core 2 CrossFire chipsets/boards currently available.
What do I think your tests show? They show that the X16 board has been better tuned for gaming performance. We've seen similar results in a lot of the tests, and I'm personally not drinking the NVIDIA Cool Aid. The FSB running at 1066 provides 8533 MB/s of bandwidth, and PCIe is 4000 MB/s bi-directional. Perhaps the FSB is aggregate, in which case you really could pull 8000 MB/s for the two X16 links... only the link between NB and SB ought to be a bottleneck as well, right?
It would be good to see some single card tests forcing PCIe to X1, X2, X4, X8, and X16 if you want another perspective on how important PCIe bandwidth really is. Doing a true look at where AGP fits into the picture is also really required for this sort of article. Obviously, SLI/CF are still going to need PCIe regardless, but is that the only advantage? Well, that and NVIDIA/ATI/Intel selling more chipsets along with all the mobo makers getting people to upgrade.» Login to reply to this |
 Freakhead_5 (70) Oct 12, 2006 - 10:54 am | Edited on Oct 12, 2006 - 10:52 am
| The conclusion should be that 99% of gamers shouldn't give one crap about pci-e 16x. Only those blowing $2000 into their cpu/gpus should give it a consideration since they obviously have money burning a hole in their pocket.
The story remains the same as b4. When new bus standards are introduced they are always way ahead of their time. And likely not necessary for years.» Login to reply to this |
 DemBones79 (226)   Oct 11, 2006 - 05:29 pm
| » Improvement how? What does this show?
Looking at the graphs and reading the numbers, I'm not seeing much of an improvement between 8x and 16x. Statistically speaking, 3fps is hardly significant. From what I understand of how the aging AGP interface (which PCIe replaced) worked, this isn't really surprising. AGP shuttled data back and forth between the system's main memory and the graphics card's on-board memory. In the event that the frame buffer and texture memory exceeded the capacity of the on-board memory, the AGP interface allowed for texture data to be shuffled to system memory, sort of like a swapfile for graphics data. This was noticeable as stuttering and lag, particularly when entering a new area and a big ol' chunk of texture data was swapped out. Texture compression went a long way toward alleviating this issue, but then games started using bigger and bigger textures (ever looked at the raw textures in UnrealEd for UT2k4?) So, then the obvious fix was to further improve the interface and increase the memory on the graphics card. The 7900 GTX used in the article has a whopping 512MB of memory on board. Some computers are still shipping with only that much memory. Let's not forget that programmers are also continuously improving their code and solutions, especially those who have been working on a console, like the Xbox, whose memory limitations require more finesse in texturing. The result is that less swapping is taking place in-game and the end result for the user is a more seemless experience.
This, in essence, renders this entire article moot. It's like testing a car's ability to make toast.
So what would make for a more informative test? Well, since the texture data needs to be sent to the gfx card in the first place, how about a comparison of load times for the same save game file from a fresh start on 8x and 16x. Or test with a card that has less on-board memory. Or run the benchmarks at such ridiculous resolutions that the framebuffer squeezes most of the texture data onto system memory. Or perhaps a combination of those conditions.
I'm not saying that the results won't be similar. It's entirely possible that there is no perceptual difference between 8x and 16x (yet). But the test design that was used shows nothing of significant value due to confounding factors.» Login to reply to this
|
» Note: You need to be logged in to write a comment!Login here, or if you don't have an account with FiringSquad, register here, it's FREE! |
  ATI Radeon 5970 Performance Preview
 After a 10-month hiatus, ATI's once again got the world's fastest graphics card. The Radeon 5970 fuses two RV870 chips onto one board for max performa... [+] (Comments) |  Left 4 Dead 2 PC Review
 Valve says Left 4 Dead 2 contains so much new content, it's worthy of a sequel rather than DLC. Is this true or false? Judge for yourself in today's r... [+] (Comments) |
Sapphire Radeon 5870 Vapor-X 1GB Review
 With its custom vapor chamber cooling+heatpipes and factory OC'ing, Sapphire's 5870 Vapor-X is targeted towards gamers looking for a 5870 card with a ... [+] (Comments) | Call of Duty: Modern Warfare 2 PC Review
 With no dedicated servers, no lean, and 18-player cap for multi, does Modern Warfare 2 for PC live up to its predecessors? Most of the reviews online ... [+] (Comments) |
Phenom II Gets A New Revision: 125W AMD Phenom II X4 965 Performance Preview
 Promising lower power consumption, lower temps, and most importantly for enthusiasts, more OC'ing, AMD is back with a new CPU revision for the Phenom ... [+] (Comments) | Dragon Age Origins Review
 |
AMD Athlon II X3 435/Athlon II X2 240e Performance Preview
 Today AMD is introducing 8 new Athlon II CPUs intended to service different segments of the budget CPU market. For HTPC users, new 45W dual, triple, a... [+] (Comments) | Shattered Horizon Review
 FutureMark, well known for their popular 3DMark benchmarks, is venturing into new territory with Shattered Horizon. This multiplayer shooter is perhap... [+] (Comments) |
| EVGA P55 FTW Review
 Looking for a good P55 motherboard to OC your CPU beyond 4GHz? If so, you may want to check out EVGA's P55 FTW. With its extra ATX12V connector, this ... [+] (Comments) | Borderlands PC Review
 Is it an RPG or is it an FPS? Borderlands blends the best elements of both in one entertaining package. Vandy has spent the past week playing the PC v... [+] (Comments) |
ATI Radeon HD 5770/5750 Performance Preview
 With prices ranging from $109-$159, ATI's Radeon 5700 series of cards bring DX11 gaming to mainstream price points and usher in new levels of energy e... [+] (Comments) | Operation Flashpoint: Dragon Rising Review
 While it's not the true sequel to Operation Flashpoint, Dragon Rising is billed as a modern tactical sim just like its predecessor. Does it live up to... [+] (Comments) |
Batman: Arkham Asylum PhysX Features and Performance
 One eye candy feature PC users can enjoy over the console edition of Batman: AA is PhysX. Rocksteady's PhysX implementation is more than just tearing ... [+] (Comments) | Batman: Arkham Asylum PC Review
 Already a smash hit on consoles, the PC version of Batman: Arkham Asylum sports better graphics and support for NVIDIA technologies PhysX and 3D Visio... [+] (Comments) |
ATI Radeon 5850 Performance Preview
 Not everyone's got $400 to spend on a shiny new Radeon 5870 card, which is why it can be argued that ATI's Radeon 5850 is the more relevant GPU for a ... [+] (Comments) | Resident Evil 5 PC Review
 Sporting a new mercenaries mode with more enemies on screen, higher resolution DX10 graphics, and 3D Vision support, Resident Evil 5 is definitely bes... [+] (Comments) |
| More Hardware » | More Games » | Interviews » |
| |
• This Month
• October 1 - 31, 2009
• September 1 - 30, 2009
• August 1 - 31, 2009
• July 1 - 31, 2009
• June 1 - 30, 2009
• May 1 - 31, 2009
• April 1 - 30, 2009
• March 1 - 31, 2009
• February 1 - 28, 2009
• January 1 - 31, 2009
• December 1 - 31, 2008
• November 1 - 30, 2008
|
|
|
Digg
Del.icio.us
Reddit
SU
furl
Previous news article
