In terms of bandwidth, PCI-E 2.1 with 8 lanes would be the same as PCI-E 1.0 at 16 lanes. However, due to the higher clock of PCI-E 2.1, there should be latency improvements to give it a small edge.
I just wanted to say thank you for including the 8800 GT in your benchmark charts. Even though its dated hardware, including it in your comparisons illustrates the punch that the newer hardware has and assists in decision making for people like myself looking to upgrade from their current setup.
Both Matlab PCT and Jacket is made on top of CUDA so it won't help with testing AMD cards. ViennaCL or ArrayFire which are both OpenCL can be used for both Nvidia and AMD
Actually we picked up ArrayFire, but it's one of our failed benchmarks as the ArrayFire benchmark collection executes too quickly and was proving to be rather insensitive to GPU performance. Not that it couldn't be the basis of an interesting benchmark, but the included benchmarks don't cut the mustard for our needs. We don't write our own benchmarks (technical skill aside, in-house benchmarks raise questions of independent verification), so we can only work with what we have.
And actually I did some research into Accelereyes' products ahead of our review, and I caught notice that Jacket 2.0 supports OpenCL We don't have that or MATLAB so I can't speak of its performance but I will leave the door open. I don't have any experience with MATLAB (it's not heavily used in pure CompSci), but if any of you do and are willing to lend us your expertise, I definitely agree it could make for an interesting benchmark.
Hi Ryan, my attention was just drawn to this thread. I'm the head of ViennaCL and looking forward to assist you with setting up an OpenCL benchmark suite. Feel free to contact me for details (my email address can be found on the ViennaCL webpage).
That looks like 64MB in under a third of a second, or more than 192MB/s. Looks like my Neal Stephenon-esque highly-secure crypto-tastic data-haven-under-a-mountain will need a sweet gaming rig in it.
To be honest, there is no point in rushing for motherboard with Gen3. slots. It isn't worth it. Of course manufacturers will use even bigger letters in marketing slogans to convince people that they need Gen3 right now. But simple truth is that only people running storage of PCI-Ex add-on RAID controllers will notice boost in performance. Because those cards can use additional bandwidth even using older Gen2 tech. Boost is not massive, but certainly noticeable. For gaming right now (and foreseeable future) Gen3 PCI-Ex is as useful as 'snooze' button on a smoke alarm. Even most modern and powerful VGAs barely saturating with data Gen2 x8 slot (no matter in single mode or SLI/QF-ed to the max). And there is Gen4 on the horizon... I wonder what for right now, hardware lagging so far behind standards...
You just said PCIE 3.0 isn't worth it but then go on to mention conditions where it WOULD be worth it. So clearly it is worth it for some, including people who do GPU compute. Secondly if someone buys a 3.0 board right now and dont make use of the extra bandwidth they could well use in the future with a GPU upgrade, and not have to buy a new board because they're stuck with PCIE 2.0, so thats a very limited imagination you have. Dont know why anyone would moan about something being more future proof.
Interfaces have seldomly been limiting and have always been used as a marketing argument. Falling for this was dumb in the 90's (and probably earlier) and it still is.
However, getting the upgrade for free is nothing to complain about. It's rather short-sighted, actually. Stoarge, GPU-Compute, Thunderbolt and multi-GPU gaming all benefit from the upgrade. And considering modern game engines like BF3 where data is continously streamed to the GPU (as storing everything there would simply be too much) there's the possibility the interface will matter even more for single cards in the future.
What about the situation with fewer-than-maximum (16 per card) CPI-e lane boards ? Most people use 16-lane boards and some of them opt for SLI/Crossfire. In this case they are limited to 8 lanes per card which allegedly is a noticable (measurable) bottleneck. With doubled throughput of PCI-e 3.0 8/8 lane split will no longer be a bottleneck and those people won't have to shell out heaps of money for an "enthusiast" board and CPU either.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
14 Comments
Back to Article
palladium - Thursday, December 22, 2011 - link
I doubt there would be a significant difference in gaming between PCIe gen 1 and gen 3. For now anyway.descendency - Thursday, December 22, 2011 - link
I may be wrong, but wouldn't a PCI 2.1 x8 be the same as a PCI 1.0 x16?Kevin G - Thursday, December 22, 2011 - link
In terms of bandwidth, PCI-E 2.1 with 8 lanes would be the same as PCI-E 1.0 at 16 lanes. However, due to the higher clock of PCI-E 2.1, there should be latency improvements to give it a small edge.tw99 - Thursday, December 22, 2011 - link
I just wanted to say thank you for including the 8800 GT in your benchmark charts. Even though its dated hardware, including it in your comparisons illustrates the punch that the newer hardware has and assists in decision making for people like myself looking to upgrade from their current setup.Drazick - Thursday, December 22, 2011 - link
Hi Anand,Why wouldn't you use MATLAB and MATLAB + JACKET for GPGPU testings?
Moreover, it would be great if you could add MATLAB for your application performance test bed.
Thanks.
Cairista - Friday, December 23, 2011 - link
I second this! (registered just to say this)MySchizoBuddy - Sunday, December 25, 2011 - link
Both Matlab PCT and Jacket is made on top of CUDA so it won't help with testing AMD cards.ViennaCL or ArrayFire which are both OpenCL can be used for both Nvidia and AMD
Ryan Smith - Sunday, December 25, 2011 - link
Actually we picked up ArrayFire, but it's one of our failed benchmarks as the ArrayFire benchmark collection executes too quickly and was proving to be rather insensitive to GPU performance. Not that it couldn't be the basis of an interesting benchmark, but the included benchmarks don't cut the mustard for our needs. We don't write our own benchmarks (technical skill aside, in-house benchmarks raise questions of independent verification), so we can only work with what we have.And actually I did some research into Accelereyes' products ahead of our review, and I caught notice that Jacket 2.0 supports OpenCL We don't have that or MATLAB so I can't speak of its performance but I will leave the door open. I don't have any experience with MATLAB (it's not heavily used in pure CompSci), but if any of you do and are willing to lend us your expertise, I definitely agree it could make for an interesting benchmark.
koarl0815 - Monday, January 2, 2012 - link
Hi Ryan,my attention was just drawn to this thread. I'm the head of ViennaCL and looking forward to assist you with setting up an OpenCL benchmark suite. Feel free to contact me for details (my email address can be found on the ViennaCL webpage).
twotwotwo - Thursday, December 22, 2011 - link
That looks like 64MB in under a third of a second, or more than 192MB/s. Looks like my Neal Stephenon-esque highly-secure crypto-tastic data-haven-under-a-mountain will need a sweet gaming rig in it.ypsylon - Friday, December 23, 2011 - link
To be honest, there is no point in rushing for motherboard with Gen3. slots. It isn't worth it. Of course manufacturers will use even bigger letters in marketing slogans to convince people that they need Gen3 right now. But simple truth is that only people running storage of PCI-Ex add-on RAID controllers will notice boost in performance. Because those cards can use additional bandwidth even using older Gen2 tech. Boost is not massive, but certainly noticeable. For gaming right now (and foreseeable future) Gen3 PCI-Ex is as useful as 'snooze' button on a smoke alarm. Even most modern and powerful VGAs barely saturating with data Gen2 x8 slot (no matter in single mode or SLI/QF-ed to the max). And there is Gen4 on the horizon... I wonder what for right now, hardware lagging so far behind standards...B3an - Saturday, December 24, 2011 - link
You just said PCIE 3.0 isn't worth it but then go on to mention conditions where it WOULD be worth it. So clearly it is worth it for some, including people who do GPU compute. Secondly if someone buys a 3.0 board right now and dont make use of the extra bandwidth they could well use in the future with a GPU upgrade, and not have to buy a new board because they're stuck with PCIE 2.0, so thats a very limited imagination you have. Dont know why anyone would moan about something being more future proof.MrSpadge - Wednesday, December 28, 2011 - link
Interfaces have seldomly been limiting and have always been used as a marketing argument. Falling for this was dumb in the 90's (and probably earlier) and it still is.However, getting the upgrade for free is nothing to complain about. It's rather short-sighted, actually. Stoarge, GPU-Compute, Thunderbolt and multi-GPU gaming all benefit from the upgrade. And considering modern game engines like BF3 where data is continously streamed to the GPU (as storing everything there would simply be too much) there's the possibility the interface will matter even more for single cards in the future.
MrS
Arnulf - Friday, December 23, 2011 - link
What about the situation with fewer-than-maximum (16 per card) CPI-e lane boards ? Most people use 16-lane boards and some of them opt for SLI/Crossfire. In this case they are limited to 8 lanes per card which allegedly is a noticable (measurable) bottleneck. With doubled throughput of PCI-e 3.0 8/8 lane split will no longer be a bottleneck and those people won't have to shell out heaps of money for an "enthusiast" board and CPU either.