Performance of llama.cpp with Vulkan #10879
Replies: 147 comments 250 replies
-
|
AMD FirePro W8100
|
Beta Was this translation helpful? Give feedback.
-
|
With the latest updates:
|
Beta Was this translation helpful? Give feedback.
-
|
With FA:
|
Beta Was this translation helpful? Give feedback.
-
|
AMD RX 470
|
Beta Was this translation helpful? Give feedback.
-
|
With the latest updates:
|
Beta Was this translation helpful? Give feedback.
-
|
With FA:
|
Beta Was this translation helpful? Give feedback.
-
|
i got the mining edition 8 gb ran slightly better
build: e288693 (6242) |
Beta Was this translation helpful? Give feedback.
-
|
ubuntu 24.04, vulkan and cuda installed from official APT packages.
build: 4da69d1 (4351) vs CUDA on the same build/setup
build: 4da69d1 (4351) |
Beta Was this translation helpful? Give feedback.
-
|
Macbook Air M2 on Asahi Linux ggml_vulkan: Found 1 Vulkan devices:
|
Beta Was this translation helpful? Give feedback.
-
|
For the record I think this is slow on the HoneyKrisp side rather than llama.cpp |
Beta Was this translation helpful? Give feedback.
-
|
Can you share how you got vulkan to build on Asahi? I can't seem to get cmake to notice it. |
Beta Was this translation helpful? Give feedback.
-
|
Spoke too soon, got it working! |
Beta Was this translation helpful? Give feedback.
-
|
Gentoo Linux on ROG Ally (2023) Ryzen Z1 Extreme ggml_vulkan: Found 1 Vulkan devices:
|
Beta Was this translation helpful? Give feedback.
-
|
ggml_vulkan: Found 4 Vulkan devices:
|
Beta Was this translation helpful? Give feedback.
-
|
Cool setup! Could you also post the result of 1, 2 and 3 7900 XTX GPUs? You can use only the first GPU with |
Beta Was this translation helpful? Give feedback.
-
|
|
Beta Was this translation helpful? Give feedback.
-
|
For this multi GPU case getting Vulkan to support #6017 pipeline parallelism might help improve the prompt processing speed. |
Beta Was this translation helpful? Give feedback.
-
|
@netrunnereve I updated the commit id in all my result. |
Beta Was this translation helpful? Give feedback.
-
|
build: 0d52a69 (4439) NVIDIA GeForce RTX 3090 (NVIDIA)
AMD Radeon RX 6800 XT (RADV NAVI21) (radv)
AMD Radeon (TM) Pro VII (RADV VEGA20) (radv)
Intel(R) Arc(tm) A770 Graphics (DG2) (Intel open-source Mesa driver)
|
Beta Was this translation helpful? Give feedback.
-
|
@netrunnereve Some of the tg results here are a little low, I think they might be debug builds. The cmake step (at least on Linux) might require |
Beta Was this translation helpful? Give feedback.
-
|
I've added My release numbers for the RX 470 are basically identical to the ones I posted earlier without the flag.
|
Beta Was this translation helpful? Give feedback.
-
|
Maybe not in your case, but some other results are suspiciously low in tg (for example the RTX 3080) |
Beta Was this translation helpful? Give feedback.
-
|
Build: 8d59d91 (4450)
Lack of proper Xe coopmat support in the ANV driver is a setback honestly.
edit: retested both with the default batch size. |
Beta Was this translation helpful? Give feedback.
-
I put my A770 into a Windows PC and gave Intel GPA and vtune a shot: GPA just crashes most of the time, I couldn't get it to trace anything useful. vtune works, but does not support Vulkan. It just shows some high-level metrics in that case, not really useful sadly. |
Beta Was this translation helpful? Give feedback.
-
I did build it with cmake with build type Release. |
Beta Was this translation helpful? Give feedback.
-
|
In that case it's something else, cause it should be performing similarly to my A770. I suspect the mesa version, there was something in newer mesa versions that slowed down tg on Intel. |
Beta Was this translation helpful? Give feedback.
-
|
A750 has 448 CUs, A770 has 512 CUs I think. Personally, I am not worried about tg. I am worried about pp here. The gemm batch quickly saturates my GPU. |
Beta Was this translation helpful? Give feedback.
-
|
@0cc4m https://gitlab.freedesktop.org/mesa/mesa/-/issues/12585 |
Beta Was this translation helpful? Give feedback.
-
|
Here's something exotic: An AMD FirePro S10000 dual GPU from 2012 with 2x 3GB GDDR5. build: 914a82d (4452)
|
Beta Was this translation helpful? Give feedback.
-
|
Very interesting, and looks like it's pretty close to the W8100 in tg despite being a dual GPU card. Your backend scales pretty well with layer splitting which is why I find it worthwhile to run my RX470 and W8100 together (I end up getting results that are close to the average of both cards).
|
Beta Was this translation helpful? Give feedback.
-
|
Latest arch with For the sake of consistency I run every bit in a script and also build every target from scratch (for some reason kill -STOP -1
timeout 240s $COMMAND
kill -CONT -1
ggml_vulkan: Found 1 Vulkan devices:
ggml_vulkan: 0 = Intel(R) Iris(R) Xe Graphics (TGL GT2) (Intel open-source Mesa driver) | uma: 1 | fp16: 1 | warp size: 32 | matrix cores: none
build: ff3fcab (4459)
This bit seems to underutilise both GPU and CPU in real conditions based on
|
Beta Was this translation helpful? Give feedback.
-
|
Unless you reduce the number of GPU layers, threads and openblas/non-openblas is not gonna make any difference. Try it with ngl 0, then only prompt processing is accelerated using Vulkan, the rest runs on CPU. This is often a good setting for integrated GPUs. |
Beta Was this translation helpful? Give feedback.
-
|
That's something I didn't think about, with
build: ba8a1f9 (4460)
|
Beta Was this translation helpful? Give feedback.
-
|
It seems latest patches has improved the results a bit: ggml_vulkan: Found 1 Vulkan devices:
|
Beta Was this translation helpful? Give feedback.
-
|
A few months later and I get:
build: f3a4b16 (5568) I run it on Linux (Arch with lama.cpp-vulkan-git package compiled by GCC 15). From my tests, only Vulkan backend ( I'm curious why I cannot go over 6 t/s. Is this an issue with the newer llama.cpp version or with my OS configuration? |
Beta Was this translation helpful? Give feedback.
-
|
Intel ARC A770 on Windows:
build: ba8a1f9 (4460) |
Beta Was this translation helpful? Give feedback.
-
Single GPU VulkanRadeon Instinct MI25 ggml_vulkan: 0 = AMD Radeon Instinct MI25 (RADV VEGA10) (radv) | uma: 0 | fp16: 1 | warp size: 64 | matrix cores: none
build: 2739a71 (4461) Radeon PRO VII ggml_vulkan: 0 = AMD Radeon Pro VII (RADV VEGA20) (radv) | uma: 0 | fp16: 1 | warp size: 64 | matrix cores: none
build: 2739a71 (4461) Multi GPU Vulkanggml_vulkan: 0 = AMD Radeon Pro VII (RADV VEGA20) (radv) | uma: 0 | fp16: 1 | warp size: 64 | matrix cores: none
build: 2739a71 (4461) ggml_vulkan: 0 = AMD Radeon Pro VII (RADV VEGA20) (radv) | uma: 0 | fp16: 1 | warp size: 64 | matrix cores: none
build: 2739a71 (4461) Single GPU RocmDevice 0: AMD Radeon Instinct MI25, compute capability 9.0, VMM: no
build: 2739a71 (4461) Device 0: AMD Radeon Pro VII, compute capability 9.0, VMM: no
build: 2739a71 (4461) Multi GPU RocmDevice 0: AMD Radeon Pro VII, compute capability 9.0, VMM: no
build: 2739a71 (4461) Layer split
build: 2739a71 (4461) Row split
build: 2739a71 (4461) Single GPU speed is decent, but multi GPU trails Rocm by a wide margin, especially with large models due to the lack of row split. |
Beta Was this translation helpful? Give feedback.
-
|
What is the power profile for this MI25? Mine is 110W but its running slower than yours on git from today. |
Beta Was this translation helpful? Give feedback.
-
|
Mine defaults to 220w. |
Beta Was this translation helpful? Give feedback.
-
|
AMD Radeon RX 5700 XT on Arch using mesa-git and setting a higher GPU power limit compared to the stock card.
I also think it could be interesting adding the flash attention results to the scoreboard (even if the support for it still isn't as mature as CUDA's).
|
Beta Was this translation helpful? Give feedback.
-
|
There is no Vulkan flash attention support (except with coopmat2 on very new nvidia drivers). What you're measuring here is a CPU fallback. |
Beta Was this translation helpful? Give feedback.
-
|
I see, I was sure about the CPU fallback but didn't know there was no flash attention support at all. |
Beta Was this translation helpful? Give feedback.
-
|
I tried but there's nothing after 1 hrs , ok, might be 40 mins... Anyway I run the llama_cli for a sample eval...
Meanwhile OpenBLAS |
Beta Was this translation helpful? Give feedback.
-
|
Even at below 1t/s |
Beta Was this translation helpful? Give feedback.
-
|
Truth is ... (0.79 tokens per second), 3788.63 ms / 3 tokens So it's not even...it just slower... |
Beta Was this translation helpful? Give feedback.
-
|
Significant improvements visible (now close to 2/3 of IPEX LLM) .\llama-bench.exe -m ..\llama-2-7b.Q4_0.gguf -ngl 99
build: d1d8241 (6193) |
Beta Was this translation helpful? Give feedback.
-
|
AMD Radeon R9 Fury
build: c8c4495 (5820)
build: 1a99c2d (6213) Sadly, driver for RX 470 mining was not successfully installed so gpu-z only reported that the card only support OpenGL and no vulkan |
Beta Was this translation helpful? Give feedback.
-
build: e288693 (6242) |
Beta Was this translation helpful? Give feedback.
-
|
That's more like it 😄. Looks like something went wrong with the FA run through, maybe you ran out of vram? |
Beta Was this translation helpful? Give feedback.
-
|
vram should still be within range. here's the verbose run, now it seems worse for tg https://gist.github.com/pebaryan/b5f1fb8d9effed0c8ceddb8d336ac4ca |
Beta Was this translation helpful? Give feedback.
-
|
Hmm if the speed is so inconsistent I wonder if the driver's using gtt like in #13765. You can check this by running radeontop and seeing how much vram and gtt is being used. |
Beta Was this translation helpful? Give feedback.
-
|
in the begining, the values are but then it changes to it seems you're right, the GTT usage increased without FA, the values are then drops to |
Beta Was this translation helpful? Give feedback.
-
|
Tested on SAPPHIRE AMD Radeon RX 7900 XT GAMING 20G Vulkan:
ROCm:
|
Beta Was this translation helpful? Give feedback.
-
|
Tested on HP NVIDIA RTX A2000 12GB Cuda:
Vulkan:
|
Beta Was this translation helpful? Give feedback.
-
|
ggml_vulkan: Found 1 Vulkan devices:
|
Beta Was this translation helpful? Give feedback.
-
|
ggml_vulkan: 0 = NVIDIA GeForce RTX 2080 Ti (NVIDIA) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 0 | matrix cores: KHR_coopmat
|
Beta Was this translation helpful? Give feedback.
-
|
Here are updated values from my hardware. Some pretty good improvements. Vega20 (Radeon VII, MI50, MI60) is looking pretty competitive now with Vulkan: ggml_vulkan: 0 = AMD Radeon (TM) Pro VII (RADV VEGA20) (radv) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 64 | shared memory: 65536 | int dot: 1 | matrix cores: none
ggml_vulkan: 0 = AMD Radeon RX 6800 XT (RADV NAVI21) (radv) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 65536 | int dot: 1 | matrix cores: none
ggml_vulkan: 0 = Tesla P40 (NVIDIA) | uma: 0 | fp16: 0 | bf16: 0 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: none
ggml_vulkan: 0 = NVIDIA GeForce RTX 3090 (NVIDIA) | uma: 0 | fp16: 1 | bf16: 1 | warp size: 32 | shared memory: 49152 | int dot: 1 | matrix cores: NV_coopmat2
ggml_vulkan: 0 = Intel(R) Arc(tm) A770 Graphics (DG2) (Intel open-source Mesa driver) | uma: 0 | fp16: 1 | bf16: 1 | warp size: 32 | shared memory: 65536 | int dot: 1 | matrix cores: none
(SYCL just segfaults on me, so no comparison) |
Beta Was this translation helpful? Give feedback.
-
|
Hey @0cc4m - is there a way of running llama-bench with only specified backends like I see in your tables here? Right now in order to get "clean" runs I am separately compiling Vulkan and ROCm and it's kind of a pain. |
Beta Was this translation helpful? Give feedback.
-
|
I also compiled them separately, but in theory you can compile both backends into one binary with |
Beta Was this translation helpful? Give feedback.
-
Radeon Pro WX 5100ggml_vulkan: 0 = AMD Radeon (TM) Pro WX 5100 Graphics (RADV POLARIS10) (radv) | uma: 0 | fp16: 0 | bf16: 0 | warp size: 64 | shared memory: 65536 | int dot: 0 | matrix cores: none
Radeon Pro V620ggml_vulkan: 1 = AMD Radeon PRO V620 (RADV NAVI21) (radv) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 65536 | int dot: 1 | matrix cores: none
build: d1e2adb (6382) System specs: Xeon W-2150B, 256GB DDR4 RAM (4x64GB, quad channel), Ubuntu 24.04 LXC in Proxmox 9 VE host |
Beta Was this translation helpful? Give feedback.
-
|
What clock speed is the WX 5100 running at during inference? I have a feeling that you're being power limited really badly. |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for the heads-up. Not sure how best to monitor... using Interestingly Edit: looking at the BIOS files at TechPowerUp they all list similar power limits. Am I missing something? |
Beta Was this translation helpful? Give feedback.
-
|
That's probably normal, IIRC AMD cards just report the power draw of the main chip, so parts of the full TDP are covered by memory and other components that are not in that number. |
Beta Was this translation helpful? Give feedback.
-
|
Yeah that 47W number is correct, and it's what you'll see in As for the BIOS whoever created it doesn't know what they're doing as they allow for a 50% power limit increase in software which is impossible with a 49A TDC, so turning it up won't do much unless you modify the BIOS. |
Beta Was this translation helpful? Give feedback.
-
|
Significant TG gains on my 1660 Ti. Now TG is 30% faster, but prompt processing lags by 10% (with
1 GHz:
build: b43556e (6424) |
Beta Was this translation helpful? Give feedback.
-
|
Can you check whether #15729 helps with your prompt processing perf? |
Beta Was this translation helpful? Give feedback.
-
|
I didn't see an improvement. (Maybe, a little bit in non-FA, but that's it.)
build: 616bca3 (6351) |
Beta Was this translation helpful? Give feedback.
-
|
This is an i7-8700T in a Lenovo ThinkCentre M920q ggml_vulkan: 0 = Intel(R) UHD Graphics 630 (CFL GT2) (Intel open-source Mesa driver) | uma: 1 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 65536 | int dot: 0 | matrix cores: none
build: c4df49a (6401) And here's the CPU-only view (using 6 of 12 threads): load_backend: loaded RPC backend from /root/cpu/build/bin/libggml-rpc.so
build: c4df49a (6401) |
Beta Was this translation helpful? Give feedback.
-
AMD Radeon RX 580 (4GB)
build: c97b5e5 (6405) on |
Beta Was this translation helpful? Give feedback.
-
|
Here's a 4GB RX 470 with Mesa 25.2.2 and
|
Beta Was this translation helpful? Give feedback.
-
|
Tried nogttspill, for me GTT issue still persists, commented in #13765. |
Beta Was this translation helpful? Give feedback.
-
|
Radeon 660M (Ryzen 5 6600H), 8GB UMA buffer size, dual channel DDR5 4800 MT/s.
|
Beta Was this translation helpful? Give feedback.
-
|
RX 6900 XT ggml_vulkan: Found 1 Vulkan devices:
build: a972fae (6428) |
Beta Was this translation helpful? Give feedback.
-
|
with framework desktop @140W. ggml_vulkan: Found 1 Vulkan devices:
build: 03b92fc (6127) |
Beta Was this translation helpful? Give feedback.
-
|
Look the wrap size is wrong with the build .... it is a RDNA3.5 it may be best with 32. |
Beta Was this translation helpful? Give feedback.
and others
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
This is similar to the Apple Silicon benchmark thread, but for Vulkan! Many improvements have been made to the Vulkan backend and I think it's good to consolidate and discuss our results here.
We'll be testing the Llama 2 7B model like the other thread to keep things consistent, and use Q4_0 as it's simple to compute and small enough to fit on a 4GB GPU. You can download it here.
Instructions
Either run the commands below or download one of our Vulkan releases. If you have multiple GPUs please run the test on a single GPU using
-sm none -mg YOUR_GPU_NUMBERunless the model is too big to fit in VRAM.Share your llama-bench results along with the git hash and Vulkan info string in the comments. Feel free to try other models and compare backends, but only valid runs will be placed on the scoreboard.
If multiple entries are posted for the same setup I'll prioritize newer commits with substantial Vulkan updates, otherwise I'll pick the one with the highest overall score at my discretion. Performance may vary depending on driver, operating system, board manufacturer, etc. even if the chip is the same. For integrated graphics note that the memory speed and number of channels will greatly affect your inference speed!
Vulkan Scoreboard for Llama 2 7B, Q4_0 (no FA)
Vulkan Scoreboard for Llama 2 7B, Q4_0 (with FA)
Beta Was this translation helpful? Give feedback.
All reactions