From 25e98e6f60dd52b40940f79e05925e8cfde6cfaf Mon Sep 17 00:00:00 2001 From: junq <22017000+QiJune@users.noreply.github.com> Date: Wed, 26 Nov 2025 15:03:09 +0800 Subject: [PATCH 1/2] fix math formula rendering issues Signed-off-by: junq <22017000+QiJune@users.noreply.github.com> --- .../run-benchmark-with-trtllm-serve.md | 20 +++++++++---------- ...loyment-guide-for-deepseek-r1-on-trtllm.md | 20 +++++++++---------- .../deployment-guide-for-gpt-oss-on-trtllm.md | 20 +++++++++---------- ...oyment-guide-for-llama3.3-70b-on-trtllm.md | 20 +++++++++---------- ...oyment-guide-for-llama4-scout-on-trtllm.md | 20 +++++++++---------- 5 files changed, 50 insertions(+), 50 deletions(-) diff --git a/docs/source/commands/trtllm-serve/run-benchmark-with-trtllm-serve.md b/docs/source/commands/trtllm-serve/run-benchmark-with-trtllm-serve.md index 76713213708c..00780df98686 100644 --- a/docs/source/commands/trtllm-serve/run-benchmark-with-trtllm-serve.md +++ b/docs/source/commands/trtllm-serve/run-benchmark-with-trtllm-serve.md @@ -161,34 +161,34 @@ P99 E2EL (ms): 1643.44 For a single request, ITLs are the time intervals between tokens, while TPOT is the average of those intervals: -```math +$$ \text{TPOT (1\ request)} = \text{Avg(ITL)} = \frac{\text{E2E\ latency} - \text{TTFT}}{\text{\#Output\ Tokens} - 1} -``` +$$ Across different requests, **average TPOT** is the mean of each request's TPOT (all requests weighted equally), while **average ITL** is token-weighted (all tokens weighted equally): -```math +$$ \text{Avg TPOT (N requests)} = \frac{\text{TPOT}_1 + \text{TPOT}_2 + \cdots + \text{TPOT}_N}{N} -``` +$$ -```math +$$ \text{Avg ITL (N requests)} = \frac{\text{Sum of all ITLs across requests}}{\text{\#Output Tokens across requests}} -``` +$$ #### End-to-End (E2E) Latency * The typical total time from when a request is submitted until the final token of the response is received. #### Total Token Throughput * The combined rate at which the system processes both input (prompt) tokens and output (generated) tokens. -```math +$$ \text{Total\ TPS} = \frac{\text{\#Input\ Tokens}+\text{\#Output\ Tokens}}{T_{last} - T_{first}} -``` +$$ #### Tokens Per Second (TPS) or Output Token Throughput * how many output tokens the system generates each second. -```math +$$ \text{TPS} = \frac{\text{\#Output\ Tokens}}{T_{last} - T_{first}} -``` +$$ ### Request Time Breakdown diff --git a/docs/source/deployment-guide/deployment-guide-for-deepseek-r1-on-trtllm.md b/docs/source/deployment-guide/deployment-guide-for-deepseek-r1-on-trtllm.md index 8b0b89ec885c..30e410cd6130 100644 --- a/docs/source/deployment-guide/deployment-guide-for-deepseek-r1-on-trtllm.md +++ b/docs/source/deployment-guide/deployment-guide-for-deepseek-r1-on-trtllm.md @@ -399,31 +399,31 @@ P99 E2EL (ms): [result] For a single request, ITLs are the time intervals between tokens, while TPOT is the average of those intervals: -```math +$$ \text{TPOT (1\ request)} = \text{Avg(ITL)} = \frac{\text{E2E\ latency} - \text{TTFT}}{\text{\#Output\ Tokens} - 1} -``` +$$ Across different requests, **average TPOT** is the mean of each request's TPOT (all requests weighted equally), while **average ITL** is token-weighted (all tokens weighted equally): -```math +$$ \text{Avg TPOT (N requests)} = \frac{\text{TPOT}_1 + \text{TPOT}_2 + \cdots + \text{TPOT}_N}{N} -``` +$$ -```math +$$ \text{Avg ITL (N requests)} = \frac{\text{Sum of all ITLs across requests}}{\text{\#Output Tokens across requests}} -``` +$$ #### End-to-End (E2E) Latency * The typical total time from when a request is submitted until the final token of the response is received. #### Total Token Throughput * The combined rate at which the system processes both input (prompt) tokens and output (generated) tokens. -```math +$$ \text{Total\ TPS} = \frac{\text{\#Input\ Tokens}+\text{\#Output\ Tokens}}{T_{last} - T_{first}} -``` +$$ #### Tokens Per Second (TPS) or Output Token Throughput * how many output tokens the system generates each second. -```math +$$ \text{TPS} = \frac{\text{\#Output\ Tokens}}{T_{last} - T_{first}} -``` +$$ diff --git a/docs/source/deployment-guide/deployment-guide-for-gpt-oss-on-trtllm.md b/docs/source/deployment-guide/deployment-guide-for-gpt-oss-on-trtllm.md index 7c8c55112760..ee4df1d05436 100644 --- a/docs/source/deployment-guide/deployment-guide-for-gpt-oss-on-trtllm.md +++ b/docs/source/deployment-guide/deployment-guide-for-gpt-oss-on-trtllm.md @@ -349,31 +349,31 @@ P99 E2EL (ms): [result] For a single request, ITLs are the time intervals between tokens, while TPOT is the average of those intervals: -```math +$$ \text{TPOT (1\ request)} = \text{Avg(ITL)} = \frac{\text{E2E\ latency} - \text{TTFT}}{\text{\#Output\ Tokens} - 1} -``` +$$ Across different requests, **average TPOT** is the mean of each request's TPOT (all requests weighted equally), while **average ITL** is token-weighted (all tokens weighted equally): -```math +$$ \text{Avg TPOT (N requests)} = \frac{\text{TPOT}_1 + \text{TPOT}_2 + \cdots + \text{TPOT}_N}{N} -``` +$$ -```math +$$ \text{Avg ITL (N requests)} = \frac{\text{Sum of all ITLs across requests}}{\text{\#Output Tokens across requests}} -``` +$$ #### End-to-End (E2E) Latency * The typical total time from when a request is submitted until the final token of the response is received. #### Total Token Throughput * The combined rate at which the system processes both input (prompt) tokens and output (generated) tokens. -```math +$$ \text{Total\ TPS} = \frac{\text{\#Input\ Tokens}+\text{\#Output\ Tokens}}{T_{last} - T_{first}} -``` +$$ #### Tokens Per Second (TPS) or Output Token Throughput * how many output tokens the system generates each second. -```math +$$ \text{TPS} = \frac{\text{\#Output\ Tokens}}{T_{last} - T_{first}} -``` +$$ diff --git a/docs/source/deployment-guide/deployment-guide-for-llama3.3-70b-on-trtllm.md b/docs/source/deployment-guide/deployment-guide-for-llama3.3-70b-on-trtllm.md index 6c16d1c2ca55..272df2629a39 100644 --- a/docs/source/deployment-guide/deployment-guide-for-llama3.3-70b-on-trtllm.md +++ b/docs/source/deployment-guide/deployment-guide-for-llama3.3-70b-on-trtllm.md @@ -354,31 +354,31 @@ P99 E2EL (ms): [result] For a single request, ITLs are the time intervals between tokens, while TPOT is the average of those intervals: -```math +$$ \text{TPOT (1\ request)} = \text{Avg(ITL)} = \frac{\text{E2E\ latency} - \text{TTFT}}{\text{\#Output\ Tokens} - 1} -``` +$$ Across different requests, **average TPOT** is the mean of each request's TPOT (all requests weighted equally), while **average ITL** is token-weighted (all tokens weighted equally): -```math +$$ \text{Avg TPOT (N requests)} = \frac{\text{TPOT}_1 + \text{TPOT}_2 + \cdots + \text{TPOT}_N}{N} -``` +$$ -```math +$$ \text{Avg ITL (N requests)} = \frac{\text{Sum of all ITLs across requests}}{\text{\#Output Tokens across requests}} -``` +$$ #### End-to-End (E2E) Latency * The typical total time from when a request is submitted until the final token of the response is received. #### Total Token Throughput * The combined rate at which the system processes both input (prompt) tokens and output (generated) tokens. -```math +$$ \text{Total\ TPS} = \frac{\text{\#Input\ Tokens}+\text{\#Output\ Tokens}}{T_{last} - T_{first}} -``` +$$ #### Tokens Per Second (TPS) or Output Token Throughput * how many output tokens the system generates each second. -```math +$$ \text{TPS} = \frac{\text{\#Output\ Tokens}}{T_{last} - T_{first}} -``` +$$ diff --git a/docs/source/deployment-guide/deployment-guide-for-llama4-scout-on-trtllm.md b/docs/source/deployment-guide/deployment-guide-for-llama4-scout-on-trtllm.md index 9fb6b6165af2..0da1e5365465 100644 --- a/docs/source/deployment-guide/deployment-guide-for-llama4-scout-on-trtllm.md +++ b/docs/source/deployment-guide/deployment-guide-for-llama4-scout-on-trtllm.md @@ -346,31 +346,31 @@ P99 E2EL (ms): [result] For a single request, ITLs are the time intervals between tokens, while TPOT is the average of those intervals: -```math +$$ \text{TPOT (1\ request)} = \text{Avg(ITL)} = \frac{\text{E2E\ latency} - \text{TTFT}}{\text{\#Output\ Tokens} - 1} -``` +$$ Across different requests, **average TPOT** is the mean of each request's TPOT (all requests weighted equally), while **average ITL** is token-weighted (all tokens weighted equally): -```math +$$ \text{Avg TPOT (N requests)} = \frac{\text{TPOT}_1 + \text{TPOT}_2 + \cdots + \text{TPOT}_N}{N} -``` +$$ -```math +$$ \text{Avg ITL (N requests)} = \frac{\text{Sum of all ITLs across requests}}{\text{\#Output Tokens across requests}} -``` +$$ #### End-to-End (E2E) Latency * The typical total time from when a request is submitted until the final token of the response is received. #### Total Token Throughput * The combined rate at which the system processes both input (prompt) tokens and output (generated) tokens. -```math +$$ \text{Total\ TPS} = \frac{\text{\#Input\ Tokens}+\text{\#Output\ Tokens}}{T_{last} - T_{first}} -``` +$$ #### Tokens Per Second (TPS) or Output Token Throughput * how many output tokens the system generates each second. -```math +$$ \text{TPS} = \frac{\text{\#Output\ Tokens}}{T_{last} - T_{first}} -``` +$$ From f9c2fe4a590595bb13d0b703446457516550df15 Mon Sep 17 00:00:00 2001 From: junq <22017000+QiJune@users.noreply.github.com> Date: Wed, 26 Nov 2025 16:04:42 +0800 Subject: [PATCH 2/2] update Signed-off-by: junq <22017000+QiJune@users.noreply.github.com> --- .../trtllm-serve/run-benchmark-with-trtllm-serve.md | 10 ++++++---- .../deployment-guide-for-deepseek-r1-on-trtllm.md | 10 ++++++---- .../deployment-guide-for-gpt-oss-on-trtllm.md | 10 ++++++---- .../deployment-guide-for-llama3.3-70b-on-trtllm.md | 10 ++++++---- .../deployment-guide-for-llama4-scout-on-trtllm.md | 10 ++++++---- 5 files changed, 30 insertions(+), 20 deletions(-) diff --git a/docs/source/commands/trtllm-serve/run-benchmark-with-trtllm-serve.md b/docs/source/commands/trtllm-serve/run-benchmark-with-trtllm-serve.md index 00780df98686..b9aed01cf7fd 100644 --- a/docs/source/commands/trtllm-serve/run-benchmark-with-trtllm-serve.md +++ b/docs/source/commands/trtllm-serve/run-benchmark-with-trtllm-serve.md @@ -162,7 +162,7 @@ P99 E2EL (ms): 1643.44 For a single request, ITLs are the time intervals between tokens, while TPOT is the average of those intervals: $$ -\text{TPOT (1\ request)} = \text{Avg(ITL)} = \frac{\text{E2E\ latency} - \text{TTFT}}{\text{\#Output\ Tokens} - 1} +\text{TPOT (1 request)} = \text{Avg(ITL)} = \frac{\text{E2E latency} - \text{TTFT}}{\text{#Output Tokens} - 1} $$ Across different requests, **average TPOT** is the mean of each request's TPOT (all requests weighted equally), while **average ITL** is token-weighted (all tokens weighted equally): @@ -172,7 +172,7 @@ $$ $$ $$ -\text{Avg ITL (N requests)} = \frac{\text{Sum of all ITLs across requests}}{\text{\#Output Tokens across requests}} +\text{Avg ITL (N requests)} = \frac{\text{Sum of all ITLs across requests}}{\text{#Output Tokens across requests}} $$ #### End-to-End (E2E) Latency @@ -180,14 +180,16 @@ $$ #### Total Token Throughput * The combined rate at which the system processes both input (prompt) tokens and output (generated) tokens. + $$ -\text{Total\ TPS} = \frac{\text{\#Input\ Tokens}+\text{\#Output\ Tokens}}{T_{last} - T_{first}} +\text{Total TPS} = \frac{\text{#Input Tokens}+\text{#Output Tokens}}{T_{last} - T_{first}} $$ #### Tokens Per Second (TPS) or Output Token Throughput * how many output tokens the system generates each second. + $$ -\text{TPS} = \frac{\text{\#Output\ Tokens}}{T_{last} - T_{first}} +\text{TPS} = \frac{\text{#Output Tokens}}{T_{last} - T_{first}} $$ ### Request Time Breakdown diff --git a/docs/source/deployment-guide/deployment-guide-for-deepseek-r1-on-trtllm.md b/docs/source/deployment-guide/deployment-guide-for-deepseek-r1-on-trtllm.md index 30e410cd6130..5ad959eaeb76 100644 --- a/docs/source/deployment-guide/deployment-guide-for-deepseek-r1-on-trtllm.md +++ b/docs/source/deployment-guide/deployment-guide-for-deepseek-r1-on-trtllm.md @@ -400,7 +400,7 @@ P99 E2EL (ms): [result] For a single request, ITLs are the time intervals between tokens, while TPOT is the average of those intervals: $$ -\text{TPOT (1\ request)} = \text{Avg(ITL)} = \frac{\text{E2E\ latency} - \text{TTFT}}{\text{\#Output\ Tokens} - 1} +\text{TPOT (1 request)} = \text{Avg(ITL)} = \frac{\text{E2E latency} - \text{TTFT}}{\text{#Output Tokens} - 1} $$ Across different requests, **average TPOT** is the mean of each request's TPOT (all requests weighted equally), while **average ITL** is token-weighted (all tokens weighted equally): @@ -410,7 +410,7 @@ $$ $$ $$ -\text{Avg ITL (N requests)} = \frac{\text{Sum of all ITLs across requests}}{\text{\#Output Tokens across requests}} +\text{Avg ITL (N requests)} = \frac{\text{Sum of all ITLs across requests}}{\text{#Output Tokens across requests}} $$ #### End-to-End (E2E) Latency @@ -418,12 +418,14 @@ $$ #### Total Token Throughput * The combined rate at which the system processes both input (prompt) tokens and output (generated) tokens. + $$ -\text{Total\ TPS} = \frac{\text{\#Input\ Tokens}+\text{\#Output\ Tokens}}{T_{last} - T_{first}} +\text{Total TPS} = \frac{\text{#Input Tokens}+\text{#Output Tokens}}{T_{last} - T_{first}} $$ #### Tokens Per Second (TPS) or Output Token Throughput * how many output tokens the system generates each second. + $$ -\text{TPS} = \frac{\text{\#Output\ Tokens}}{T_{last} - T_{first}} +\text{TPS} = \frac{\text{#Output Tokens}}{T_{last} - T_{first}} $$ diff --git a/docs/source/deployment-guide/deployment-guide-for-gpt-oss-on-trtllm.md b/docs/source/deployment-guide/deployment-guide-for-gpt-oss-on-trtllm.md index ee4df1d05436..9378eec095a3 100644 --- a/docs/source/deployment-guide/deployment-guide-for-gpt-oss-on-trtllm.md +++ b/docs/source/deployment-guide/deployment-guide-for-gpt-oss-on-trtllm.md @@ -350,7 +350,7 @@ P99 E2EL (ms): [result] For a single request, ITLs are the time intervals between tokens, while TPOT is the average of those intervals: $$ -\text{TPOT (1\ request)} = \text{Avg(ITL)} = \frac{\text{E2E\ latency} - \text{TTFT}}{\text{\#Output\ Tokens} - 1} +\text{TPOT (1 request)} = \text{Avg(ITL)} = \frac{\text{E2E latency} - \text{TTFT}}{\text{#Output Tokens} - 1} $$ Across different requests, **average TPOT** is the mean of each request's TPOT (all requests weighted equally), while **average ITL** is token-weighted (all tokens weighted equally): @@ -360,7 +360,7 @@ $$ $$ $$ -\text{Avg ITL (N requests)} = \frac{\text{Sum of all ITLs across requests}}{\text{\#Output Tokens across requests}} +\text{Avg ITL (N requests)} = \frac{\text{Sum of all ITLs across requests}}{\text{#Output Tokens across requests}} $$ #### End-to-End (E2E) Latency @@ -368,12 +368,14 @@ $$ #### Total Token Throughput * The combined rate at which the system processes both input (prompt) tokens and output (generated) tokens. + $$ -\text{Total\ TPS} = \frac{\text{\#Input\ Tokens}+\text{\#Output\ Tokens}}{T_{last} - T_{first}} +\text{Total TPS} = \frac{\text{#Input Tokens}+\text{#Output Tokens}}{T_{last} - T_{first}} $$ #### Tokens Per Second (TPS) or Output Token Throughput * how many output tokens the system generates each second. + $$ -\text{TPS} = \frac{\text{\#Output\ Tokens}}{T_{last} - T_{first}} +\text{TPS} = \frac{\text{#Output Tokens}}{T_{last} - T_{first}} $$ diff --git a/docs/source/deployment-guide/deployment-guide-for-llama3.3-70b-on-trtllm.md b/docs/source/deployment-guide/deployment-guide-for-llama3.3-70b-on-trtllm.md index 272df2629a39..e7709a17aa5f 100644 --- a/docs/source/deployment-guide/deployment-guide-for-llama3.3-70b-on-trtllm.md +++ b/docs/source/deployment-guide/deployment-guide-for-llama3.3-70b-on-trtllm.md @@ -355,7 +355,7 @@ P99 E2EL (ms): [result] For a single request, ITLs are the time intervals between tokens, while TPOT is the average of those intervals: $$ -\text{TPOT (1\ request)} = \text{Avg(ITL)} = \frac{\text{E2E\ latency} - \text{TTFT}}{\text{\#Output\ Tokens} - 1} +\text{TPOT (1 request)} = \text{Avg(ITL)} = \frac{\text{E2E latency} - \text{TTFT}}{\text{#Output Tokens} - 1} $$ Across different requests, **average TPOT** is the mean of each request's TPOT (all requests weighted equally), while **average ITL** is token-weighted (all tokens weighted equally): @@ -365,7 +365,7 @@ $$ $$ $$ -\text{Avg ITL (N requests)} = \frac{\text{Sum of all ITLs across requests}}{\text{\#Output Tokens across requests}} +\text{Avg ITL (N requests)} = \frac{\text{Sum of all ITLs across requests}}{\text{#Output Tokens across requests}} $$ #### End-to-End (E2E) Latency @@ -373,12 +373,14 @@ $$ #### Total Token Throughput * The combined rate at which the system processes both input (prompt) tokens and output (generated) tokens. + $$ -\text{Total\ TPS} = \frac{\text{\#Input\ Tokens}+\text{\#Output\ Tokens}}{T_{last} - T_{first}} +\text{Total TPS} = \frac{\text{#Input Tokens}+\text{#Output Tokens}}{T_{last} - T_{first}} $$ #### Tokens Per Second (TPS) or Output Token Throughput * how many output tokens the system generates each second. + $$ -\text{TPS} = \frac{\text{\#Output\ Tokens}}{T_{last} - T_{first}} +\text{TPS} = \frac{\text{#Output Tokens}}{T_{last} - T_{first}} $$ diff --git a/docs/source/deployment-guide/deployment-guide-for-llama4-scout-on-trtllm.md b/docs/source/deployment-guide/deployment-guide-for-llama4-scout-on-trtllm.md index 0da1e5365465..60c2b306c75a 100644 --- a/docs/source/deployment-guide/deployment-guide-for-llama4-scout-on-trtllm.md +++ b/docs/source/deployment-guide/deployment-guide-for-llama4-scout-on-trtllm.md @@ -347,7 +347,7 @@ P99 E2EL (ms): [result] For a single request, ITLs are the time intervals between tokens, while TPOT is the average of those intervals: $$ -\text{TPOT (1\ request)} = \text{Avg(ITL)} = \frac{\text{E2E\ latency} - \text{TTFT}}{\text{\#Output\ Tokens} - 1} +\text{TPOT (1 request)} = \text{Avg(ITL)} = \frac{\text{E2E latency} - \text{TTFT}}{\text{#Output Tokens} - 1} $$ Across different requests, **average TPOT** is the mean of each request's TPOT (all requests weighted equally), while **average ITL** is token-weighted (all tokens weighted equally): @@ -357,7 +357,7 @@ $$ $$ $$ -\text{Avg ITL (N requests)} = \frac{\text{Sum of all ITLs across requests}}{\text{\#Output Tokens across requests}} +\text{Avg ITL (N requests)} = \frac{\text{Sum of all ITLs across requests}}{\text{#Output Tokens across requests}} $$ #### End-to-End (E2E) Latency @@ -365,12 +365,14 @@ $$ #### Total Token Throughput * The combined rate at which the system processes both input (prompt) tokens and output (generated) tokens. + $$ -\text{Total\ TPS} = \frac{\text{\#Input\ Tokens}+\text{\#Output\ Tokens}}{T_{last} - T_{first}} +\text{Total TPS} = \frac{\text{#Input Tokens}+\text{#Output Tokens}}{T_{last} - T_{first}} $$ #### Tokens Per Second (TPS) or Output Token Throughput * how many output tokens the system generates each second. + $$ -\text{TPS} = \frac{\text{\#Output\ Tokens}}{T_{last} - T_{first}} +\text{TPS} = \frac{\text{#Output Tokens}}{T_{last} - T_{first}} $$