mirror of
https://github.com/kvcache-ai/ktransformers.git
synced 2026-04-24 00:09:04 +00:00
* refactor: move legacy code to archive/ directory - Moved ktransformers, csrc, third_party, merge_tensors to archive/ - Moved build scripts and configurations to archive/ - Kept kt-kernel, KT-SFT, doc, and README files in root - Preserved complete git history for all moved files * refactor: restructure repository to focus on kt-kernel and KT-SFT modules * fix README * fix README * fix README * fix README * docs: add performance benchmarks to kt-kernel section Add comprehensive performance data for kt-kernel to match KT-SFT's presentation: - AMX kernel optimization: 21.3 TFLOPS (3.9× faster than PyTorch) - Prefill phase: up to 20× speedup vs baseline - Decode phase: up to 4× speedup - NUMA optimization: up to 63% throughput improvement - Multi-GPU (8×L20): 227.85 tokens/s total throughput with DeepSeek-R1 FP8 Source: https://lmsys.org/blog/2025-10-22-KTransformers/ This provides users with concrete performance metrics for both core modules, making it easier to understand the capabilities of each component. * refactor: improve kt-kernel performance data with specific hardware and models Replace generic performance descriptions with concrete benchmarks: - Specify exact hardware: 8×L20 GPU + Xeon Gold 6454S, Single/Dual-socket Xeon + AMX - Include specific models: DeepSeek-R1-0528 (FP8), DeepSeek-V3 (671B) - Show detailed metrics: total throughput, output throughput, concurrency details - Match KT-SFT presentation style for consistency This provides users with actionable performance data they can use to evaluate hardware requirements and expected performance for their use cases. * fix README * docs: clean up performance table and improve formatting * add pic for README * refactor: simplify .gitmodules and backup legacy submodules - Remove 7 legacy submodules from root .gitmodules (archive/third_party/*) - Keep only 2 active submodules for kt-kernel (llama.cpp, pybind11) - Backup complete .gitmodules to archive/.gitmodules - Add documentation in archive/README.md for researchers who need legacy submodules This reduces initial clone size by ~500MB and avoids downloading unused dependencies. * refactor: move doc/ back to root directory Keep documentation in root for easier access and maintenance. * refactor: consolidate all images to doc/assets/ - Move kt-kernel/assets/heterogeneous_computing.png to doc/assets/ - Remove KT-SFT/assets/ (images already in doc/assets/) - Update KT-SFT/README.md image references to ../doc/assets/ - Eliminates ~7.9MB image duplication - Centralizes all documentation assets in one location * fix pic path for README
58 lines
2.3 KiB
C++
58 lines
2.3 KiB
C++
#include <future>
|
|
#include "common.hpp"
|
|
|
|
int main(int argc, char* argv[]) {
|
|
init(argc, argv);
|
|
spdlog::set_level(spdlog::level::debug);
|
|
config.gpu_cache_config->total_kvcache_pages = 12;
|
|
auto kvc2 = kvc2::create_kvc2(config);
|
|
|
|
// #pragma omp parallel for
|
|
for (size_t ti = 0; ti < 2; ti++) {
|
|
SPDLOG_WARN("Test {}", ti);
|
|
auto [kcache, vcache] = kvc2->get_kvcache();
|
|
std::mt19937 gen(ti + 123);
|
|
size_t total_page = 10;
|
|
TokenLength total_length = total_page * config.num_token_per_page;
|
|
auto tokens = random_ids(total_length, gen);
|
|
TokenLength prompt_length = 3 * config.num_token_per_page;
|
|
auto k1 = random_kvcache(total_page, gen);
|
|
auto v1 = random_kvcache(total_page, gen);
|
|
|
|
{
|
|
std::promise<std::shared_ptr<DoubleCacheHandleInterface>> p;
|
|
kvc2->lookup_to_gpu_async(test_model_name, test_quant_type, tokens.data(), prompt_length, total_length,
|
|
[&p](std::shared_ptr<DoubleCacheHandleInterface> h) { p.set_value(h); });
|
|
auto fut = p.get_future();
|
|
fut.wait();
|
|
auto h = fut.get();
|
|
assert(h->matched_length() % config.num_token_per_page == 0);
|
|
size_t matched_block = h->matched_length() / config.num_token_per_page;
|
|
auto block_idx = h->get_gpu_block_idx();
|
|
cmp_handle_gpu(block_idx, kcache, vcache, k1, v1, matched_block);
|
|
for (size_t at = matched_block; at < block_idx.size(); at++) {
|
|
copy_cpu_gpu(block_idx, kcache, vcache, k1, v1, at);
|
|
}
|
|
h->append_tokens(tokens.data(), total_length);
|
|
cmp_handle_gpu(block_idx, kcache, vcache, k1, v1, total_page);
|
|
}
|
|
|
|
{
|
|
std::promise<std::shared_ptr<DoubleCacheHandleInterface>> p;
|
|
kvc2->lookup_to_gpu_async(test_model_name, test_quant_type, tokens.data(), total_length, total_length,
|
|
[&p](std::shared_ptr<DoubleCacheHandleInterface> h) { p.set_value(h); });
|
|
auto fut = p.get_future();
|
|
fut.wait();
|
|
auto h = fut.get();
|
|
assert(h->matched_length() == total_length);
|
|
size_t matched_block = h->matched_length() / config.num_token_per_page;
|
|
auto block_idx = h->get_gpu_block_idx();
|
|
cmp_handle_gpu(block_idx, kcache, vcache, k1, v1, matched_block);
|
|
}
|
|
}
|
|
kvc2->save();
|
|
SPDLOG_CRITICAL("All Test Passed: {}", argv[0]);
|
|
|
|
return 0;
|
|
}
|