[SYCL][Intel GPU] Long Term Features & Issues Tracking

[airMeng]

Feel free to drop a note, let's know if you have any feature request or bugs (even unconfirmed)

• Multi-card Support. Issue #5282, PR #5806
• Multi-batch Support #5272 Low performance with Sycl Backend #5480
• CI test error for more than one GPU is detected and used.
  Current code returns all SYCL devices, including CPU, GPU (level-zero, opencl), FPGA. SYCL only support GPU. So when CI test on other devices, it will be fault.
• Support no-mmap parameter in other application.
  There is known issue of SYCL: memcpy() from host (mmap) to device will hang in same cases. It's not resolved now. A work around solution is no use mmap. I have handled it in llama-bench (add --mmap parameter). We need add to more applications in examples.
• Clean code for warning and unused macro and variable.
  Suggest to handle it after multiple-card is finished. Lots of such unused code will be useful for multiple-card feature.
• Support SYCL build for Nvidia and AMD targets #5357
• Improve first token performance.

Also let's know if you have taken any tasks here.

cc @NeoZhangJianyu @luoyu-intel @abhilash1910

[NeoZhangJianyu]

I'd like handle: Multi-card Support, CI test error for more than one GPU is detected and used.

[abhilash1910]

For code cleaning and sanitization of compiler runtime I will be adding patch post the previous changes
Also sycl for other vendor build targets.

[DatCaptainHorse]

Would like to see support for SOTA 2-bit quantized models (GGML_TYPE_IQ2_XXS, GGML_TYPE_IQ2_XS, GGML_TYPE_IQ3_XXS).

Been trying to do this myself for past hour or two, using dpct isn't as trouble-free as it makes you think.

[characharm]

before airMeng:sycl_fix_max_alloc_size

I don't know if these numbers can be considered when the model produces useless output.

Using device 0 (Intel(R) Arc(TM) A770 Graphics) as main device

model	size	params	backend	ngl	test	t/s
llama 13B Q5_K - Medium	8.60 GiB	13.02 B	GPU BLAS	99	pp 512	430.52 ± 38.83
llama 13B Q5_K - Medium	8.60 GiB	13.02 B	GPU BLAS	99	tg 128	16.14 ± 0.13

model	size	params	backend	ngl	test	t/s
llama 7B Q6_K	5.53 GiB	7.24 B	GPU BLAS	99	pp 512	734.90 ± 149.16
llama 7B Q6_K	5.53 GiB	7.24 B	GPU BLAS	99	tg 128	22.00 ± 0.17

after

model	size	params	backend	ngl	test	t/s
llama 13B Q5_K - Medium	8.60 GiB	13.02 B	SYCL	99	pp 512	392.43 ± 33.81
llama 13B Q5_K - Medium	8.60 GiB	13.02 B	SYCL	99	tg 128	9.75 ± 0.06

model	size	params	backend	ngl	test	t/s
llama 7B Q6_K	5.53 GiB	7.24 B	SYCL	99	pp 512	703.17 ± 122.82
llama 7B Q6_K	5.53 GiB	7.24 B	SYCL	99	tg 128	19.07 ± 0.15

vulkan latest

model	size	params	backend	ngl	test	t/s
llama 13B Q5_K - Medium	8.60 GiB	13.02 B	Vulkan	99	pp 512	33.99 ± 0.53
llama 13B Q5_K - Medium	8.60 GiB	13.02 B	Vulkan	99	tg 128	8.04 ± 0.01

model	size	params	backend	ngl	test	t/s
llama 7B Q6_K	5.53 GiB	7.24 B	Vulkan	99	pp 512	74.76 ± 3.49
llama 7B Q6_K	5.53 GiB	7.24 B	Vulkan	99	tg 128	21.26 ± 0.10

[qnixsynapse]

I wonder why build fails with -DLLAMA_SYCL_F16=ON for my Intel Arc 750..

I think this GPU support f16.

But anyways getting 16 tokens/sec for q4_k_m for 7B with all layers on GPU.

With batch bench, mitral -7B Q4_K_M:

PP	TG	B	N_KV	T_PP s	S_PP t/s	T_TG s	S_TG t/s	T s	S t/s
128	128	1	256	0.497	257.31	8.129	15.75	8.626	29.68
128	128	2	512	0.539	475.29	43.357	5.90	43.896	11.66
128	128	4	1024	0.799	641.06	30.104	17.01	30.903	33.14
128	128	8	2048	1.546	662.36	30.739	33.31	32.285	63.44
128	128	16	4096	3.261	627.93	32.109	63.78	35.371	115.80
128	256	1	384	0.338	378.75	15.331	16.70	15.669	24.51
128	256	2	768	0.452	566.45	86.797	5.90	87.249	8.80
128	256	4	1536	0.759	674.89	60.522	16.92	61.281	25.06
128	256	8	3072	1.548	661.63	61.703	33.19	63.251	48.57

[NeoZhangJianyu]

@akarshanbiswas
It should be fixed by PR: #5411.
The FP16 build is added to CI. There won't be build issue of FP16 in the feature.

By the way, if you have more issues, please create/update in the "issue" field for it, instead of "discussion" field.

[qnixsynapse]

Oh okay. The title of this discussion has "issues tracking" so I mentioned that issue here. :)

[airMeng]

@akarshanbiswas Could you report the full performance here like the above comments since we don't have A750 on our side, so a performance baseline on A750 would be quite helpful? Thank you for your effort to help improving SYCL backend.

You can find the standard performance measurements here

[qnixsynapse]

@airMeng Thank you. I was able to test a little and I have updated my comment here.

[kunger97]

I would like to know if there are plans to support quantization types that are not currently supported like iq3, iq4?

[characharm]

With the recent changes, the model nous-hermes-2-34b-2.69 has seen a significant speedup. From the unusable 3-4 tokens per second, it now reaches 7-8

detect 1 SYCL GPUs: [0,] with Max compute units:512

model	size	params	backend	ngl	test	t/s
llama 30B Q2_K - Medium	10.76 GiB	34.39 B	SYCL	99	pp 512	159.76 ± 8.53
llama 30B Q2_K - Medium	10.76 GiB	34.39 B	SYCL	99	tg 128	8.52 ± 0.02

build: 21b0867 (2345)

for comparison

Vulkan0: Intel(R) Arc(TM) A770 Graphics | uma: 0 | fp16: 1 | warp size: 32

model	size	params	backend	ngl	test	t/s
llama 30B Q2_K - Medium	10.76 GiB	34.39 B	Vulkan	99	pp 512	33.27 ± 0.29
llama 30B Q2_K - Medium	10.76 GiB	34.39 B	Vulkan	99	tg 128	5.05 ± 0.05

build: 82cb31e (2348)

I hope other quantization methods will also see an improvement, for now, they more or less perform similarly to Vulkan.

[qnixsynapse]

I have a question. During prompt processing or generation, the llama.cpp's SYCL backend seems to use only one of the (I am assuming XMX) engines of my GPU. Although they are tagged 'unknown' in intel_gpu_top. Does anyone knows why? And is it possible to parallelize across all of them?

On Linux, I can't even monitor neither the VRAM usage, nor the temps which is surprising because it is a while since this GPU was launched. My all hopes for the new Xe driver.

[airMeng]

I have a question. During prompt processing or generation, the llama.cpp's SYCL backend seems to use only one of the (I am assuming XMX) engines of my GPU. Although they are tagged 'unknown' in intel_gpu_top. Does anyone knows why? And is it possible to parallelize across all of them?

On Linux, I can't even monitor neither the VRAM usage, nor the temps which is surprising because it is a while since this GPU was launched. My all hopes for the new Xe driver.

@akarshanbiswas can you try https://github.com/intel/xpumanager to monitor the usage?

[qnixsynapse]

It is not yet available for Arch Linux. I think I will have to compile it from source. Also, it is for datacenter GPUs. Will it support Intel Arc?

p.s I didn't find any sysfs interface for monitoring temps or vram usage. Will this work with currently stable mainline kernel?

[airMeng]

will it support Intel Arc

yes it is.

I didn't find any sysfs interface for monitoring temps or vram usage

you can get temps and vram usage here https://github.com/intel/xpumanager/blob/fdcb817c0dfadf9423da4dfe6d99fa712f62f5b6/doc/smi_user_guide.md?plain=1#L637

[qnixsynapse]

I ended up successfully building the package and it shows this output(with root):

+-----------------------------+--------------------------------------------------------------------+
| Device ID                   | 0                                                                  |
+-----------------------------+--------------------------------------------------------------------+
| GPU Utilization (%)         | N/A                                                                |
| EU Array Active (%)         | N/A                                                                |
| EU Array Stall (%)          | N/A                                                                |
| EU Array Idle (%)           | N/A                                                                |
|                             |                                                                    |
| Compute Engine Util (%)     | N/A                                                                |
| Render Engine Util (%)      | Engine 0: 0                                                        |
| Media Engine Util (%)       | N/A                                                                |
| Decoder Engine Util (%)     | Engine 0: 0, Engine 1: 0                                           |
| Encoder Engine Util (%)     | Engine 0: 0, Engine 1: 0                                           |
| Copy Engine Util (%)        | Engine 0: 0                                                        |
| Media EM Engine Util (%)    | Engine 0: 0, Engine 1: 0                                           |
| 3D Engine Util (%)          | N/A                                                                |
+-----------------------------+--------------------------------------------------------------------+
| Reset                       | N/A                                                                |
| Programming Errors          | N/A                                                                |
| Driver Errors               | N/A                                                                |
| Cache Errors Correctable    | N/A                                                                |
| Cache Errors Uncorrectable  | N/A                                                                |
| Mem Errors Correctable      | N/A                                                                |
| Mem Errors Uncorrectable    | N/A                                                                |
+-----------------------------+--------------------------------------------------------------------+
| GPU Power (W)               | 37                                                                 |
| GPU Frequency (MHz)         | 2400                                                               |
| Media Engine Freq (MHz)     | N/A                                                                |
| GPU Core Temperature (C)    | N/A                                                                |
| GPU Memory Temperature (C)  | N/A                                                                |
| GPU Memory Read (kB/s)      | N/A                                                                |
| GPU Memory Write (kB/s)     | N/A                                                                |
| GPU Memory Bandwidth (%)    | N/A                                                                |
| GPU Memory Used (MiB)       | 6804                                                               |
| GPU Memory Util (%)         | 84                                                                 |
| Xe Link Throughput (kB/s)   | N/A                                                                |
+-----------------------------+--------------------------------------------------------------------+

Got the memory usage now but temps are still N/A.

[qnixsynapse]

Adding this here. This may add to the list of todos and fixes. Credit goes to Gemini 1.5 Pro 1 million. :)

• Device-to-device memory copy across GPUs: The current implementation uses a workaround (copying data to host and then back to the other device) for device-to-device memory copies across different GPUs. (This goes along with multi card support)
  [NeoZhangJianyu] Yes, it's known limitation. More efficient implement depends on other library. This function won't impact the performance due to less call times. ✅

• Missing support for certain data types: Some data types, like GGML_TYPE_IQ4_NL, GGML_TYPE_IQ2_S, and GGML_TYPE_IQ4_XS, are not supported in the SYCL backend. Issue: [SYCL] Support newer non linear quantization #5674, PR for GGML_TYPE IQ2_S : [SYCL] iq2_s #6052, A fix (?): IQ1_S: attempt to fix SYCL #6014
  Solved in support/fix more IQ OPs #6521 ✅

• Limited use of SYCL extensions: The code could potentially benefit from utilizing more SYCL extensions for optimization. For example, the ext_oneapi_usm_device_read_only extension could be used for constant memory, and the ext_intel_free_memory extension could provide more accurate memory usage information.
  [AkarshanBiswas] ext_intel_free_memory is used but not sure about ext_oneapi_usm_device_read_only
  [NeoZhangJianyu] Already use them, like ext_intel_free_memory. ✅

• Thread safety: Some parts of the code, like the buffer type map in ggml_backend_sycl_split_buffer_type, might not be thread-safe. This could lead to issues when using multiple models or contexts simultaneously.
  [NeoZhangJianyu] Is there any test case for the issue? I don't test with multiple modes case.
  [AkarshanBiswas] Yes. Test it with continuous batching with slots > 1 and see if the driver is stable.

• Memory management: Implementing a buffer pool for SYCL similar to the CUDA version could improve memory allocation and reuse efficiency.
  [NeoZhangJianyu] The buffer pool is used in SYCL backend. ✅

• Flash attention SYCL implementation after ggml : add Flash Attention #5021

Will update when I find anything else with my limited knowledge.

[airMeng]

#6521

[piDack]

flashattention have any progress?i find huggingface intel backend have impl it https://github.com/intel/intel-extension-for-deepspeed/tree/774453156c4ad6f4b831fb985494254645f5fe1d/intel_extension_for_deepspeed/op_builder/csrc/flash_attn

[slaren]

The SYCL backend should be updated to adopt these changes in ggml-backend:

• Remove the code in ggml.c and implement the offload_op interface: backend : offload large batches to GPU #6083
• Remove the code that copies GGML_BACKEND_TYPE_CPU tensors automatically to VRAM. After adopting the previous change, this will not be necessary, all the tensors received by the SYCL backend will always be allocated in a SYCL buffer.
• Remove all usage of ggml_tensor::backend, as this will be removed in the future. To support split buffer types, use ggml_tensor::buffer to identify the storage type of the tensor instead.

[NeoZhangJianyu]

Yes, we have a plan to fix them by 2 PRs.

[airMeng]

@slaren May I know more context about offload_op? Seems only a ggml_backend_sycl_offload_op needed?

https://github.com/ggerganov/llama.cpp/pull/6217/files

[slaren]

The purpose of offload_op is to offload computation to the GPU when the batch size is large enough, even if the weights are not stored in VRAM and have to be copied. For small batch sizes the cost of copying the weights is higher than the cost of the computation, but for large enough batch sizes, it is often faster to offload the computation to the GPU since it is usually much faster than the CPU.

Previously, this was implemented by hooking into the CPU backend in ggml.c in the function ggml_compute_forward, and doing the computation in a call to ggml_sycl_compute_forward. Now, this is handled in ggml_backend_sched entirely, and the backends only need to implement the offload_op function to choose what operations they wish to run even when the weights are on system memory. The weights are copied to VRAM by ggml_backend_sched, so backends do not need to check if the tensors are in RAM and copy them, they can assume that all the tensors they receive in a call to graph_compute will always be allocated in the backend-specific buffer.

Your implementation is a good first step, but you should also remove the code of SYCL in ggml.c. To test this, you can try prompt processing without fully offloading the model. For example, llama-bench -n 0 -ngl 0. For a backend such as SYCL, it may be good to consider the type of accelerator being used, and possibly disable for slow devices. This will also allow you to remove a lot of the code in the SYCL backend that deals with copying tensors between CPU and GPU, and generally simplify this logic.

You would also need to modify llama.cpp in llama_new_context_with_model to always create an instance of the SYCL backend, even when model->n_gpu_layers == 0.

[airMeng]

@slaren for point 2 and 3, can we refer to #6170 ?

[slaren]

Sure, if you find that useful you can use it. However, there will be further refactoring of the CUDA backend in #6269.

[qnixsynapse]

Just an update here: I did not use llama.cpp for like few days because I was busy. I ran it today to test llama-3 and I found out that it hangs here everytime with every model right here:

......................................................................................
llama_new_context_with_model: n_ctx      = 512
llama_new_context_with_model: n_batch    = 512
llama_new_context_with_model: n_ubatch   = 512
llama_new_context_with_model: freq_base  = 500000.0
llama_new_context_with_model: freq_scale = 1
llama_kv_cache_init:  SYCL_Host KV buffer size =    64.00 MiB
llama_new_context_with_model: KV self size  =   64.00 MiB, K (f16):   32.00 MiB, V (f16):   32.00 MiB
llama_new_context_with_model:  SYCL_Host  output buffer size =     0.98 MiB
llama_new_context_with_model:      SYCL0 compute buffer size =   266.50 MiB
llama_new_context_with_model:  SYCL_Host compute buffer size =    45.01 MiB
llama_new_context_with_model: graph nodes  = 1062
llama_new_context_with_model: graph splits = 66

I am running with --no-mmap .

In the logs I found out:

kernel: Fence expiration time out i915-0000:03:00.0:server[5816]:14c!

Not sure if this is because of an update that I received on Arch Linux or not, not a single model is running with the same binary that it is used to run before.
I have intel-compute-runtime version 23.48.27912.11-2, on lts kernel 6.6.27, intel-oneapi-base-toolkit is 2024.0.0.49564 and level-zero-loader is 1.15.1-1.
Not opening an issue right now because I am not sure if it is a bug in llama.cpp or not.

Update: Not related to llama.cpp, JAX with intel-extension-for-openxla hangs too. (now confirmed)

Update 2: Came across this: intel/compute-runtime#497
Update 3: linux > 6.6.25 is broken (suspicious commit saving it here for bug reporting upstream.)

[qnixsynapse]

Yes. Here is the output of sycl-ls on my PC (I am still trying to debug):

[opencl:acc:0] Intel(R) FPGA Emulation Platform for OpenCL(TM), Intel(R) FPGA Emulation Device OpenCL 1.2  [2023.16.10.0.17_160000]
[opencl:cpu:1] Intel(R) OpenCL, 12th Gen Intel(R) Core(TM) i3-12100F OpenCL 3.0 (Build 0) [2023.16.10.0.17_160000]
[opencl:gpu:2] Intel(R) OpenCL Graphics, Intel(R) Arc(TM) A750 Graphics OpenCL 3.0 NEO  [23.48.27912]
[ext_oneapi_level_zero:gpu:0] Intel(R) Level-Zero, Intel(R) Arc(TM) A750 Graphics 1.3 [1.3.27912]

I have reproduced it with llama2-7b-q4 and that's why said earlier than it is not related to llama.cpp. Even JAX hangs. (with intel openxla pjrt plugin) .

[qnixsynapse]

Update: Fixed it!

There is a reason I did not report an issue at first since I was not sure of what was causing the problem. Now I know. And it is the new linux-firmware version: 20240409.1addd7dc-1 linux version > 6.6.25 .

I downgraded to version 6.6.25 and now it works.

Thinking of switching to debian on this system, I really hate updates breaking my important things around.

[airMeng]

can you help to report the issue to https://github.com/oneapi-src/level-zero? much thanks for your help!

[qnixsynapse]

No problem!
Yes, I will open an issue tomorrow.

[qnixsynapse]

Update: The regression happened on i915 kernel driver and potential commits(according to Matt):

a7ff84a6fe5a drm/i915/gt: Enable only one CCS for compute workload
726ff623869d drm/i915/gt: Do not generate the command streamer for all the CCS
c1f7ce2a11a9 drm/i915/gt: Disable HW load balancing for CCS

Quoting him, "That workaround requires that we disable load balancing on the compute engines, assign all underlying hardware compute resources to a single engine, and then hide all except for one of the engines from userspace (which is also why the "physical engines" count goes from 4 down to 1 as you noted). It's a pretty complicated and invasive hardware workaround, so it has the highest likelihood of accidentally introducing issues."

Fix is 4cfca03f7641 ("drm/i915/gt: Automate CCS Mode setting during engine resets") which will hopefully land on stable kernels by next week.

Adding these here so that people who are experiencing similar issues might find these helpful.

[airMeng]

Latest SYCL is broken due to #7640 (comment), I am looking into it and hope to fix it soon.

cc @NeoZhangJianyu @AidanBeltonS

[airMeng]

fixed in #7710

[mistrjirka]

I am not sure if it is expected or not but attempting to use Intel GPU ( Intel® Iris® Xe Graphics ) on Intel® Core™ i7-1165G7 results in much slower performance then on the CPU alone. I tried just the example prompt:

ZES_ENABLE_SYSMAN=1 ./build/bin/llama-cli -m models/Lexi-Llama-3-8B-Uncensored_Q4_K_M.gguf -p "Building a website can be done in 10 simple steps:" -n 400 -e -ngl 33 -sm none -mg 0
I get:

So around 1.6t/s for prompt eval and 3.55t/s for generation.
The GPU is used according to intel_gpu_top:

When I try it on CPU I get much better performance:
command: ZES_ENABLE_SYSMAN=1 ./build/bin/llama-cli -m models/Lexi-Llama-3-8B-Uncensored_Q4_K_M.gguf -p "Building a website can be done in 10 simple steps:" -n 400 -e -ngl 0 -sm none -mg 0

It also seems that the output from the GPU is garbled and generally wrong.

[airMeng]

hi @mistrjirka we have observed the extremely slow prompt evel issue and we are working on it. As for prompt generation, as you might know, iGPU and CPU shares the same DDR memory so the performance shall be on the same level. we are working on it too but expect not too much gains.

cc @luoyu-intel

[mistrjirka]

hi @mistrjirka we have observed the extremely slow prompt evel issue and we are working on it. As for prompt generation, as you might know, iGPU and CPU shares the same DDR memory so the performance shall be on the same level. we are working on it too but expect not too much gains.

cc @luoyu-intel

Thanks for the info and thank you for working on this.
I am honestly more concerned about the weird repeating output
Tried the prompt multiple times and the output is generally just nonsense. Sometimes it print random characters. It shows up even when using less gpu layers. That could point to some performance non related problem. I observed that it holds on Intel xe graphics and on Intel 620m. I am not sure if someone can replicate my problems on dedicated Card.

[airMeng]

we are aware of that too, seems the recent update of rope and mul_mat_id breaks the accuracy. we are working on it, sorry for any inconvenience.

[ColonelPhantom]

Hi, I'm trying to compile the SYCL backend but the compiler seems to almost get stuck on unicode-data.cpp? It's already taking 3 minutes and counting. Is this normal?

I do have an older, slower CPU (i5-8350U).

/opt/intel/oneapi/compiler/2024.0/bin/icpx -DGGML_SCHED_MAX_COPIES=4 -DGGML_USE_LLAMAFILE -DGGML_USE_SYCL -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -I/home/quinten/src/llama.sycl/. -Wno-narrowing -O3 -fsycl -L/lib -O3 -DNDEBUG -std=gnu++17 -I./ -I//opt/intel/oneapi/compiler/2024.0/include -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -march=native -MD -MT CMakeFiles/llama.dir/unicode-data.cpp.o -MF CMakeFiles/llama.dir/unicode-data.cpp.o.d -o CMakeFiles/llama.dir/unicode-data.cpp.o -c /home/quinten/src/llama.sycl/unicode-data.cpp

[mistrjirka]

llama3 8B FP16 need 18+GB memory. You could download the llama3 8B Q4 gguf format model in huggingface.

Yeah that is what I am already testing but I think the issue could be with the performance on quantized model which I cannot really test by using quantized model.

[mistrjirka]

llama3 8B FP16 need 18+GB memory. You could download the llama3 8B Q4 gguf format model in huggingface.

I looked that you have merged some performance related PR. I will test if the performance imporved in my case.

[mistrjirka]

llama3 8B FP16 need 18+GB memory. You could download the llama3 8B Q4 gguf format model in huggingface.

Tested master, due to #8286 but it did not really improve anything. There is no performance imporvement.

[ColonelPhantom]

llama3 8B FP16 need 18+GB memory. You could download the llama3 8B Q4 gguf format model in huggingface.

Yeah that is what I am already testing but I think the issue could be with the performance on quantized model which I cannot really test by using quantized model.

You could consider testing another model, e.g. Phi-3-mini which is <8 GB unquantized.

[mistrjirka]

llama3 8B FP16 need 18+GB memory. You could download the llama3 8B Q4 gguf format model in huggingface.

Yeah that is what I am already testing but I think the issue could be with the performance on quantized model which I cannot really test by using quantized model.

You could consider testing another model, e.g. Phi-3-mini which is <8 GB unquantized.

No it is not the quantization never mind. It is much slower than quantized versions. So I guess that sycl on integrated GPUs is just bad.

[qnixsynapse]

Does performance supposed to get worse when using the keys cache quantization to either q8 or q4? Normally with f16 gives 20 tokens/sec and with q_8 gives 5 tokens/sec on my Intel Arc.

Gemma 9B Q4_K_S:
with f16 k-cache:

With q8_0 k-cache:

[slaren]

That's another issue, but these copies are caused by the src1 being non contiguous. With f16 kv there is the mul_mat_vec_p021 kernel, but that doesn't work with quantized src0.

[JohannesGaessler]

Ah okay, I see what you mean. Yes, for batch size 1 that would be an issue.

[qnixsynapse]

Which kernel works with quantized src0 if you don't mind me asking?

[zhentaoyu]

It's ggml_sycl_op_mul_mat which need tensor cpy to make it contiguous. fp16 has some other specific kernels that can handle non-contiguous q. see kernel dispatcher here: https://github.com/ggerganov/llama.cpp/blob/master/ggml/src/ggml-sycl.cpp#L3661

[qnixsynapse]

I am guessing the equivalent in cuda is ggml_cuda_op_mul_mat.

I came across this:

if (src0_is_contiguous) {
            dev[id].src0_dd = split ? (char *) src0_extra->data_device[id] : (char *) src0->data;
        } else {
            dev[id].src0_dd = dev[id].src0_dd_alloc.alloc(ctx.pool(id), ggml_nbytes(src0));
        }

[joeatodd]

Hello - a heads up from the Codeplay side - we found that recent changes (#6408) introduced a lot of expensive device info queries:

        const int work_group_size = get_work_group_size(stream->get_device());

Unfortunately these queries aren't cached, and due to the way the DPCT headers are currently designed, this actually makes 15 different device info queries each time its called 🫠. @OuadiElfarouki from our side is working on a caching mechanism which should fix this. On Nvidia hardware we found this created a significant performance drop (down from ~10 T/s to ~2 T/s).

Like I say, we are working on a fix, and I am just posting for info. @airMeng @NeoZhangJianyu

Maybe these queries are cheap on Intel drivers, but I am seeing above a lot of discussions about performance regressions. Could be related? @zhentaoyu

[NeoZhangJianyu]

@joeatodd
I have same concern when see this code.
I will fix it by local cache.

[NeoZhangJianyu]

Recently, there are more PRs of SYCL backend, some miss to test the performance.
I will check one by one and fix them one by one.
Maybe some solution will be reverted if no function impact.

[NeoZhangJianyu]

I have provided similar solution in my previous PR. But it's no merged and canceled now.
I will create a new PR for it.

[NeoZhangJianyu]

Fixed by PR: #8286

[OuadiElfarouki]

Fix I've been working on : #8301

[joeatodd]

A bug that I introduced with #8644 reveals that we don't have any CI which tests static builds (-DBUILD_SHARED_LIBS=OFF). On #8667 @airMeng suggested to update one of the CI Dockerfiles to test this. I merged that PR since it was fixing a bug on master, and I am opening this comment to track/discuss how best to approach this.

I suggested a couple of approaches here. What do you think is best?

[kwaa]

Hello, here is a Feature Request: Can SYCL build be supported in llama.cpp's Nix Flakes?

I want to use it in NixOS but struggle to package oneAPI.

[airMeng]

I am not familiar with Nix, but I saw someone working on it from ollama NixOS/nixpkgs#327999 (comment), could you give a try?

[NeoZhangJianyu]

You could refer to the package method of windows release package in https://github.com/ggerganov/llama.cpp/blob/master/.github/workflows/build.yml: windows-latest-cmake-sycl.

We aren't familiar with NixOS in fact.
We welcome the experts of NixOS to join in this work. We could provide the support.

[kwaa]

I tried to package intel-llvm, but its buildbot script doesn't work (and there's almost no documentation on it).

I've also looked at MordragT's packaging approach, which feels a bit too cumbersome.

[MordragT]

Hey kwaa if you still need help setting it up, I could maybe create something with cachix if the build time was your problem.

[qnixsynapse]

I just found out that SYCL backend don't use joint_matrix, so, probably no Tensor cores or XMX.

This is confirmed by the comment here:

https://github.com/ggerganov/llama.cpp/blob/4134999e01f31256b15342b41c4de9e2477c4a6c/ggml/src/ggml-sycl/common.hpp#L59C1-L63C7

which means that backend can get much faster with proper implementation on discrete GPUs. It is not even utilizing the hardware properly! I need to forget about flash attention (the issue was closed for being stale)

We have many examples here and I am not sure how to implement this.

[airMeng]

hi @qnixsynapse for most GEMV cases, there will be no benefit from XMX. For first token GEMM, we rely on MKL which should utilize XMX already

[qnixsynapse]

first token GEMM

You mean (batched) prompt processing?

[airMeng]

either bs!=1 or seq!=1

[qnixsynapse]

Got it! Thanks.

[qnixsynapse]

Will it be preferable if we switch all the kernels to use oneDNN, not just matmul?
The API for oneDNN primitives looks very clean to me and said to have better compatibility.

[0xDEADFED5]

thank you SYCL crew, all the work is appreciated.

[qnixsynapse]

Hi. I have some questions.

Is this intentional?

We can simply call the functions with the required parameters which should be better right?

[qnixsynapse]

I think we should simplify this so that function for each operation can be independent in choosing its parameters.

[qnixsynapse]

Thoughts?
Also, CC @Rbiessy and maybe @abhilash1910

[Rbiessy]

On the first question about refactoring ggml_sycl_compute_forward it seems that func is used to be able to call ggml_sycl_set_peer_access in some cases before calling func. Assuming we want to keep this behavior I would suggest to keep func but to change the type of ggml_sycl_func_t so that it does not contain src0 and src1 arguments. They can all be deduced from the last ggml_tensor object.

On the second question about using mmq, we could help to review this PR. It has some large implications:

• We would need to verify the correctness of such a pass. Unfortunately running the tests is not always enough, internally we also run the model and verify that the output is not gibberish.
• We would need to verify the impact on performance for some models and some Nvidia and Intel devices. The PR would need to show evidence that it has a overall a positive impact. We should be able to help with that.

[qnixsynapse]

Thank you for sharing your insights. My original plan of refactoring compute_forward function was shelved because I decided to use GGML_UNUSED() macro for unused variables.

My current plan is to make this backend migrate to tensor->buffer and write optimized mmq kernels for Intel to speed up PP. But I am not an expert in writing SYCL kernels so I not having enough confidence to do so.

The screenshot shows that Q4_K, Q5_K and Q6_K kernels are broken, hence I have commented them out because we did this in the last PR which got merged:
https://github.com/ggerganov/llama.cpp/blob/64ae0655114f84f11a724bc6878c6f8f4a55560b/ggml/src/ggml-sycl/mmq.cpp#L816-L817

Also, no support for IQ_XX data types because these kernels were never updated.

Edit: According to this, Xe-HPG architecture has a register size of 32 KB. RDNA architecture, whose tiling settings I found in SYCL's mmq kernels has 128KB of register size. I am suspecting that register overflows may have been causing accuracy issues.

https://github.com/ggerganov/llama.cpp/blob/9f35e44592a7646a5803620eb6a3f0ed5ac90553/ggml/src/ggml-sycl/mmq.cpp#L1785-L1803

[qnixsynapse]

On the first question about refactoring ggml_sycl_compute_forward it seems that func is used to be able to call ggml_sycl_set_peer_access in some cases before calling func. Assuming we want to keep this behavior I would suggest to keep func but to change the type of ggml_sycl_func_t so that it does not contain src0 and src1 arguments. They can all be deduced from the last ggml_tensor object.

Actually, it checks if the src0 tensor is of split buffer type and then it calls ggml_sycl_set_peer_access. I initially thought to leave it as it is but it seems I have no choice but to refactor it.

The current code assumes that all the operation needs two src tensors (i.e src0 and src1) and because of that I find it difficult to add new OPs. Not all OPs need src0 and src1 tensors. Some need 4 src tensors such as flash attention (src0 as Q, src1 as K, src2 as V and src3 as mask), along with extra params such as bias value(for alibi) and logits softcap(Gemma models). Everything is there in node/dst tensor. The best is to check if src0 is present and is a split buffer and run ggml_sycl_set_peer_access before the switch case statement IMO.

[Mar2ck]

I'd really like to see SYCL support added to the RPC server.

[qnixsynapse]

SIGBUS {si_signo=SIGBUS, si_code=BUS_ADRERR, si_addr=0x74f11d03e020} ---

This one is helpful but incomplete. You have to rebuild the project with debug symbols by either adding -g to C++ flags / replacing --config Release with --config Debug and run it in a program called gdb.

I know it's a bit technical. I hope situation will improve once my refactoring work is done and merged which will correctly print the backend debug logs.

[rtlinux]

Thanks for the tips! Something like this...?

load_tensors: layer  23 assigned to device SYCL0
load_tensors: layer  24 assigned to device SYCL0
load_tensors: tensor 'token_embd.weight' (q8_0) (and 0 others) cannot be used with preferred buffer type CPU_AARCH64, using CPU instead
load_tensors: offloading 24 repeating layers to GPU
load_tensors: offloading output layer to GPU
load_tensors: offloaded 25/25 layers to GPU
load_tensors: RPC[192.168.1.178:50053] model buffer size =   129.39 MiB
load_tensors:   CPU_Mapped model buffer size =   137.94 MiB
load_tensors:        SYCL0 model buffer size =   244.35 MiB

Thread 1 "llama-cli" received signal SIGBUS, Bus error.
__memcpy_avx_unaligned_erms () at ../sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S:540
warning: 540    ../sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S: No such file or directory
(gdb)

backtrace:

[qnixsynapse]

0x00007ffff7a9da7f in ggml_backend_sycl_buffer_set_tensor(ggml_backend_buffer*, ggml_tensor*, void const*, unsigned 
long, unsigned long)

aligns with

mmap(NULL, 256245760, PROT_READ|PROT_WRITE, MAP_SHARED, 4, 0x156304000) = 0x74f11d03e000
--- SIGBUS {si_signo=SIGBUS, si_code=BUS_ADRERR, si_addr=0x74f11d03e020} ---

Does this happen even on llama-cli or llama-server too?

Also, proper way to get debug logs is to first set pagination off and bt full in gdb.

edit: Also check ReBar settings in your bios.

[rtlinux]

ReBar setting was indeed the culprit, it's all working now after turning it on in the bios. Thanks again for your help!

[qnixsynapse]

No problem!. ;)

[NeoZhangJianyu]

I think the internal refactor should support the outside features to user.
I suggest to move the focus to optimize the kernel performance on Intel GPU.

[qnixsynapse]

For optimization, it is best to vectorize the dequantize mul_mat kernels and caching the scales in local memory.
We will also need an optimized cpy kernels, especially q8_0 -> f32 and q4_0 -> f32 for better support of quantized KV cache feature of llama.cpp.
For more details, please see #11175 (comment)

[Rbiessy]

I am planning to work on renaming the occurrences of oneMKL Interface to oneMath following the recent renaming.

Currently oneMath is used for the Nvidia and AMD devices when SYCL is enable while the Intel devices directly use the Intel oneMKL product, see this FAQ for the difference between the two.. It would be valuable to also use oneMath for Intel devices (which will directly call Intel oneMKL) to use a common interface. I believe this will not impact performance if we use the compile time dispatcher.
One issue is that oneMath is not shipped with pre-built packages which can make building llama.cpp harder. The best solution to mitigate this is to provide 2 ways to use oneMath:

1. Use the optional installation provided by users with -DoneMath_DIR=/path/to/oneMath/install
2. Use CMake FetchContent to automatically pull and build oneMath if SYCL is enabled and oneMath cannot be found. This should not add more requirements for the user. Building oneMath BLAS domain with Intel oneMKL backends only takes 3s on my desktop.

Does anyone have a concern with this approach? In particular curious for @airMeng and @NeoZhangJianyu opinions.

[qnixsynapse]

Personally, I do not have problem with this. But we need to setup a proper CI(if not Intel, then Nvidia) and write some runtime tests here imo.

[Rbiessy]

More CI is probably good but I'm not sure how it relates to this suggestion.
I don't know if the public CI has the HW available to run some runtime tests. It's not a priority for us today but a nice to have.

[qnixsynapse]

Last time I checked, Intel CI had some issues here. oneMath reminded me of the idea that we can use already available HW here for CI. Nevermind.

[Rbiessy]

I have implemented the suggested changes in #12192

[qnixsynapse]

@Rbiessy @Alcpz @NeoZhangJianyu @airMeng , I am noticing a memory leak with each token output only when building with SYCL backend. I am unable to use vtune for the following reason on an up-to-date Fedora:

Building sep driver ...
In file included from /usr/src/kernels/6.13.5-200.fc41.x86_64/include/linux/module.h:22,
                 from lwpmudrv.c:37:
lwpmudrv.c:106:18: error: expected ‘,’ or ‘;’ before ‘INTEL_PMT’
  106 | MODULE_IMPORT_NS(INTEL_PMT);
      |                  ^~~~~~~~~
/usr/src/kernels/6.13.5-200.fc41.x86_64/include/linux/moduleparam.h:26:61: note: in definition of macro ‘__MODULE_INFO’
   26 |                 = __MODULE_INFO_PREFIX __stringify(tag) "=" info
      |                                                             ^~~~
/usr/src/kernels/6.13.5-200.fc41.x86_64/include/linux/module.h:299:33: note: in expansion of macro ‘MODULE_INFO’
  299 | #define MODULE_IMPORT_NS(ns)    MODULE_INFO(import_ns, ns)
      |                                 ^~~~~~~~~~~
lwpmudrv.c:106:1: note: in expansion of macro ‘MODULE_IMPORT_NS’
  106 | MODULE_IMPORT_NS(INTEL_PMT);
      | ^~~~~~~~~~~~~~~~
lwpmudrv.c:107:18: error: expected ‘,’ or ‘;’ before ‘INTEL_PMT_TELEMETRY’
  107 | MODULE_IMPORT_NS(INTEL_PMT_TELEMETRY);
      |                  ^~~~~~~~~~~~~~~~~~~
/usr/src/kernels/6.13.5-200.fc41.x86_64/include/linux/moduleparam.h:26:61: note: in definition of macro ‘__MODULE_INFO’
   26 |                 = __MODULE_INFO_PREFIX __stringify(tag) "=" info
      |                                                             ^~~~
/usr/src/kernels/6.13.5-200.fc41.x86_64/include/linux/module.h:299:33: note: in expansion of macro ‘MODULE_INFO’
  299 | #define MODULE_IMPORT_NS(ns)    MODULE_INFO(import_ns, ns)
      |                                 ^~~~~~~~~~~
lwpmudrv.c:107:1: note: in expansion of macro ‘MODULE_IMPORT_NS’
  107 | MODULE_IMPORT_NS(INTEL_PMT_TELEMETRY);

Memory grows 2GB per requests to the server. I tested with other backends(Vulkan) and it doesn't happen in there.

This started happening recently, I am not sure when. Can you all please help me in identifying it? If I run with valgrind, the entire thing crashes.

[Alcpz]

Yes. I am with @qnixsynapse. I felt reluctant of reverting because we get such good performance. Can we try to find where the leak happens and fix that?

[qnixsynapse]

@NeoZhangJianyu Since you are the author, so let me ask you.

Why are we adding element to the vector tensor_extras but never freeing it, even on a non Q4_0 model?

llama.cpp/ggml/src/ggml-sycl/ggml-sycl.cpp

Lines 337 to 339 in e0dbec0

	ggml_tensor_extra_gpu * extra = new ggml_tensor_extra_gpu{};
	tensor->extra = extra;
	ctx->tensor_extras.push_back(extra); //used to release it when destroy ctx.

[qnixsynapse]

Opened #12366

[NeoZhangJianyu]

Oh, it's my fault! :)
The extra is used to record the tensor state for reorder.
It will be released with the context release.
I think it may be the root cause.

Thank you to fix by #12366.

Need to free it in reset function as comment.

[qnixsynapse]

Need to free it in reset function as comment

Our backend do not support quantized KV cache properly yet. (It needs flash attention) :D
So, I am saving it for the future.

ps. It seems we do not have a reset interface:

llama.cpp/ggml/src/ggml-sycl/ggml-sycl.cpp

Lines 489 to 498 in 84d5475

	static const ggml_backend_buffer_i ggml_backend_sycl_buffer_interface = {
	/* .free_buffer = */ ggml_backend_sycl_buffer_free_buffer,
	/* .get_base = */ ggml_backend_sycl_buffer_get_base,
	/* .init_tensor = */ ggml_backend_sycl_buffer_init_tensor,
	/* .memset_tensor = */ NULL,
	/* .set_tensor = */ ggml_backend_sycl_buffer_set_tensor,
	/* .get_tensor = */ ggml_backend_sycl_buffer_get_tensor,
	/* .cpy_tensor = */ ggml_backend_sycl_buffer_cpy_tensor,
	/* .clear = */ ggml_backend_sycl_buffer_clear,
	/* .reset = */ NULL,

[zhouwg]

the technical atmosphere of this Intel sycl is very good and I really envy it. Intel is a really great IT company. I'd like to spend some time learning knowledge about Intel sycl in the future.

[NeoZhangJianyu]

Welcome to join us! :)

[zhouwg]

thanks so much for your warmly welcome and I will start as a beginner at the end of this month and setting up the dev envs on my Intel x86 desktop SoC equipped Linux machine.

[zhouwg]

I'm sorry that I couldn't start the learning&dev activity on ggml-sycl since the end of this month because it seems there is another tech approach of ggml-hexagon(llama.cpp backend for Qualcomm's Hexagon NPU), I'd like to try it regardless of success of failure.

I'll do my commitment on ggml-sycl because(1)I learnt too much from ggml-sycl on 04/2024 (2)Intel is a really great IT company.

[NeoZhangJianyu]

OK! It's good!
llama.cpp is a great platform to integrate many technology for AI.
It's easy to help people learn them. :)