Skip to content

Commit

Permalink
docs: change svg benchmark tables appearance
Browse files Browse the repository at this point in the history
Reduce number of FheUint types displayed in the integer benchmark
tables. Increase policy size and better columns fitting.
Remove link to enlarge image.
  • Loading branch information
soonum committed Mar 6, 2025
1 parent f60be3e commit 386dd73
Show file tree
Hide file tree
Showing 14 changed files with 915 additions and 1,877 deletions.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
361 changes: 115 additions & 246 deletions tfhe/docs/_static/cpu_integer_benchmark_tuniform_2m128_ciphertext.svg
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
289 changes: 94 additions & 195 deletions tfhe/docs/_static/cpu_integer_benchmark_tuniform_2m128_plaintext.svg
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
361 changes: 115 additions & 246 deletions tfhe/docs/_static/cpu_integer_benchmark_tuniform_2m64_ciphertext.svg
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
289 changes: 94 additions & 195 deletions tfhe/docs/_static/cpu_integer_benchmark_tuniform_2m64_plaintext.svg
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
2 changes: 1 addition & 1 deletion tfhe/docs/getting_started/benchmarks/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -12,4 +12,4 @@ make print_doc_bench_parameters

### Operation time over FheUint 64

![Click to enlarge](../../_static/cpu_gpu_integer_benchmark_fheuint64_tuniform_2m64_ciphertext.svg)
![](../../_static/cpu_gpu_integer_benchmark_fheuint64_tuniform_2m64_ciphertext.svg)
Original file line number Diff line number Diff line change
Expand Up @@ -14,21 +14,21 @@ The following tables benchmark the execution time of some operation sets using `

The next table shows the operation timings on CPU when all inputs are encrypted:

![Click to enlarge](../../../_static/cpu_integer_benchmark_tuniform_2m64_ciphertext.svg)
![](../../../_static/cpu_integer_benchmark_tuniform_2m64_ciphertext.svg)

The next table shows the operation timings on CPU when the left input is encrypted and the right is a clear scalar of the same size:

![Click to enlarge](../../../_static/cpu_integer_benchmark_tuniform_2m64_plaintext.svg)
![](../../../_static/cpu_integer_benchmark_tuniform_2m64_plaintext.svg)

## Pfail: $$2^{-128}$$

The next table shows the operation timings on CPU when all inputs are encrypted:

![Click to enlarge](../../../_static/cpu_integer_benchmark_tuniform_2m128_ciphertext.svg)
![](../../../_static/cpu_integer_benchmark_tuniform_2m128_ciphertext.svg)

The next table shows the operation timings on CPU when the left input is encrypted and the right is a clear scalar of the same size:

![Click to enlarge](../../../_static/cpu_integer_benchmark_tuniform_2m128_plaintext.svg)
![](../../../_static/cpu_integer_benchmark_tuniform_2m128_plaintext.svg)

All timings are based on parallelized Radix-based integer operations where each block is encrypted using the default parameters `PARAM_MESSAGE_2_CARRY_2_KS_PBS`. To ensure predictable timings, we perform operations in the `default` mode, which ensures that the input and output encoding are similar (i.e., the carries are always emptied).

Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -13,15 +13,15 @@ Note that these benchmarks use Gaussian parameters. `MB-PBS` stands for multi-bi

## P-fail: $$2^{-40}$$

![Click to enlarge](../../../_static/cpu_pbs_benchmark_tuniform_2m40.svg)
![](../../../_static/cpu_pbs_benchmark_tuniform_2m40.svg)

## P-fail: $$2^{-64}$$

![Click to enlarge](../../../_static/cpu_pbs_benchmark_tuniform_2m64.svg)
![](../../../_static/cpu_pbs_benchmark_tuniform_2m64.svg)

## P-fail: $$2^{-128}$$

![Click to enlarge](../../../_static/cpu_pbs_benchmark_tuniform_2m128.svg)
![](../../../_static/cpu_pbs_benchmark_tuniform_2m128.svg)

## Reproducing TFHE-rs benchmarks

Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -12,22 +12,22 @@ The cryptographic parameters `PARAM_GPU_MULTI_BIT_MESSAGE_2_CARRY_2_GROUP_3_KS_P
Below come the results for the execution on a single H100.
The following table shows the performance when the inputs of the benchmarked operation are encrypted:

![Click to enlarge](../../../_static/gpu_integer_benchmark_h100x1_multi_bit_tuniform_2m64_ciphertext.svg)
![](../../../_static/gpu_integer_benchmark_h100x1_multi_bit_tuniform_2m64_ciphertext.svg)

The following table shows the performance when the left input of the benchmarked operation is encrypted and the other is a clear scalar of the same size:

![Click to enlarge](../../../_static/gpu_integer_benchmark_h100x1_multi_bit_tuniform_2m64_plaintext.svg)
![](../../../_static/gpu_integer_benchmark_h100x1_multi_bit_tuniform_2m64_plaintext.svg)

## 2xH100

Below come the results for the execution on two H100's.
The following table shows the performance when the inputs of the benchmarked operation are encrypted:

![Click to enlarge](../../../_static/gpu_integer_benchmark_h100x2_multi_bit_tuniform_2m64_ciphertext.svg)
![](../../../_static/gpu_integer_benchmark_h100x2_multi_bit_tuniform_2m64_ciphertext.svg)

The following table shows the performance when the left input of the benchmarked operation is encrypted and the other is a clear scalar of the same size:

![Click to enlarge](../../../_static/gpu_integer_benchmark_h100x2_multi_bit_tuniform_2m64_plaintext.svg)
![](../../../_static/gpu_integer_benchmark_h100x2_multi_bit_tuniform_2m64_plaintext.svg)

## Reproducing TFHE-rs benchmarks

Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -8,11 +8,11 @@ All GPU benchmarks were launched on H100 GPUs, and rely on the multithreaded PBS

## P-fail: $$2^{-40}$$

![Click to enlarge](../../../_static/gpu_pbs_benchmark_tuniform_2m40.svg)
![](../../../_static/gpu_pbs_benchmark_tuniform_2m40.svg)

## P-fail: $$2^{-64}$$

![Click to enlarge](../../../_static/gpu_pbs_benchmark_tuniform_2m64.svg)
![](../../../_static/gpu_pbs_benchmark_tuniform_2m64.svg)

## Reproducing TFHE-rs benchmarks

Expand Down

0 comments on commit 386dd73

Please sign in to comment.