As you know most of but not all top performing systems on the project are Nvidia based cpus.
Now this shows up. So possiblly with its assistance we could run Nvidia petri optimized GPU apps on Amd GPU's?
It's claiming binary compatible.
Tom M
I wonder if this means we can use some of the semi older AMD gpu's with hgih FP64 rates to crunch tasks that use that, if so that could be a game changer for some peoples credits.
The 7800 xt has almost 3 times the fp32 speed of the 6700 xt. And twice the fp64 speed.
So all the GPU calculations may only take 1/3 as much time. That should speed up total production.
Keep us updated please.
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
The 7800 xt has almost 3 times the fp32 speed of the 6700 xt. And twice the fp64 speed.
So all the GPU calculations may only take 1/3 as much time. That should speed up total production.
Keep us updated please.
Tom M
RDNA3 cards have higher flops on paper only for the most part. they have a "dual-issue" capability which pretty much doubles the maximum theoretical throughput, however it relies heavily on the compiler to find dual-issue opportunities, which ends up very few in reality. real world compute performance isn't much better than RDNA2 in most cases.
don't believe the specs at face value. it's more nuanced and more of an "it depends" situation.
Unfortunately, testing through OpenCL is difficult because we’re relying on the compiler to find dual issue opportunities. We only see convincing dual issue behavior with FP32 adds, where the compiler emitted v_dual_add_f32 instructions. The mixed INT32 and FP32 addition test saw some benefit because the FP32 adds were dual issued, but could not generate VOPD instructions for INT32 due to a lack of VOPD instructions for INT32 operations. Fused multiply add, which is used to calculate a GPU’s headline TFLOPs number, saw very few dual issue instructions emitted. Both architectures can execute 16-bit operations at double rate, though that’s unrelated to RDNA 3’s new dual issue capability. Rather, 16-bit instructions benefit from a single operation issued in packed-math mode. In other major categories, throughput remains largely similar to RDNA 2.
I’m guessing RDNA 3’s dual issue mode will have limited impact. It relies heavily on the compiler to find VOPD possibilities, and compilers are frustratingly stupid at seeing very simple optimizations.
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
Impressive production. I am assuming you don't allow the info to be listed on the top 50 listing or you would be well placed there too.
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
Tom M
)
I wonder if this means we can use some of the semi older AMD gpu's with hgih FP64 rates to crunch tasks that use that, if so that could be a game changer for some peoples credits.
Everybody seems to be missing
)
Everybody seems to be missing the big disclaimer in the article.
ONLY for PTX or runtime compiled applications.
Of which . . there are none as far as I know in BOINC pantheon.
Keith Myers wrote: Everybody
)
Darn!!
I just upgraded from 6700 XT
)
I just upgraded from 6700 XT (12GB) to 7800XT (16GB). It will be interesting to see if the tasks are crunched any faster.
Guðni Már Gilbert wrote: I
)
https://www.techpowerup.com/gpu-specs/radeon-rx-6700-xt.c3695
https://www.techpowerup.com/gpu-specs/radeon-rx-7800-xt.c3839
The 7800 xt has almost 3 times the fp32 speed of the 6700 xt. And twice the fp64 speed.
So all the GPU calculations may only take 1/3 as much time. That should speed up total production.
Keep us updated please.
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
Tom M wrote:Guðni Már
)
RDNA3 cards have higher flops on paper only for the most part. they have a "dual-issue" capability which pretty much doubles the maximum theoretical throughput, however it relies heavily on the compiler to find dual-issue opportunities, which ends up very few in reality. real world compute performance isn't much better than RDNA2 in most cases.
don't believe the specs at face value. it's more nuanced and more of an "it depends" situation.
https://chipsandcheese.com/2023/01/07/microbenchmarking-amds-rdna-3-graphics-architecture/
_________________________________________________________________________
https://www.tomshardware.com/
)
https://www.tomshardware.com/pc-components/gpus/newegg-is-selling-an-old-but-unused-rx-5600-xt-gpu-for-dollar109-its-super-cheap-and-heres-how-it-stacks-up-in-2024
A golden oldy.
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
All-Sky Gravitational Wave
)
All-Sky Gravitational Wave search on O3 v1.07 () windows_x86_64
AMD W7700 (4 GPU-Tasks parallel: using app_config.xml with 0.9 CPU and 0.25 GPU.
GPU W7700
maeax wrote: All-Sky
)
Impressive production. I am assuming you don't allow the info to be listed on the top 50 listing or you would be well placed there too.
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
Tom M wrote: Impressive
)
he only just started contributing. it takes time to build RAC enough to get on the leaderboard
_________________________________________________________________________