Calculate the TPS of the CMP 70HX on local AI models
Every model in our catalogue assessed against this card at the context length and minimum quality you choose. Speed is an estimate for a single request, calculated from this card's memory bandwidth and the size of each model once compressed.
Calculated for this card
679 models in our catalogue altogether
Largest model it holds
Baichuan 1-13B
13.3B · Q3_K_M · 52.4 tok/s
Fastest model
Gemma 3 QAT 1B
258 tok/s · 1B
Which AI models can run on a CMP 70HX?
Set the inputs, read the answer
More context means more memory for the conversation cache. Speed is for a fresh conversation and does not change with this setting.
Hides models that would only fit by being compressed below this point.
337 models match
Calculating| Quantisation | Fit | ||||||
|---|---|---|---|---|---|---|---|
|
258
tok/s
219–309 |
Gemma 3 1B | 1B | Mar 2025 | 1.8 GB | 33k tokens | Q8_0 | Comfortable |
|
258
tok/s
219–309 |
Gemma 3 QAT 1B | 1B | Apr 2025 | 1.8 GB | 33k tokens | Q8_0 | Comfortable |
|
258
tok/s
155–412 · low confidence |
HGRN 1B (WT 103) ≈ | 1B | Nov 2023 | 1.8 GB | 131k tokens ? | Q8_0 | Comfortable |
|
258
tok/s
155–412 · low confidence |
LLama 3..2 Typhoon 2 1B ≈ | 1B | Dec 2024 | 1.8 GB | 131k tokens ? | Q8_0 | Comfortable |
|
258
tok/s
155–412 · low confidence |
OLMo-1B ≈ | 1B | Feb 2024 | 1.8 GB | 131k tokens ? | Q8_0 | Comfortable |
|
258
tok/s
155–412 · low confidence |
Pythia-1b ≈ | 1B | Apr 2023 | 1.8 GB | 131k tokens ? | Q8_0 | Comfortable |
|
239
tok/s
143–382 · low confidence |
OpenELM-1.1B ≈ | 1.1B | May 2024 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
234
tok/s
141–375 · low confidence |
DeciCoder-1B ≈ | 1.1B | Aug 2023 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
234
tok/s
141–375 · low confidence |
SantaCoder ≈ | 1.1B | Jan 2023 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
234
tok/s
141–375 · low confidence |
TinyLlama-1.1B (1T token checkpoint) ≈ | 1.1B | Oct 2023 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
234
tok/s
141–375 · low confidence |
TinyLlama-1.1B (3T token checkpoint) ≈ | 1.1B | Oct 2023 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
215
tok/s
129–344 · low confidence |
EXAONE 4.0 (1.2B) ≈ | 1.2B | Jul 2025 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
215
tok/s
129–344 · low confidence |
MinerU2.5 ≈ | 1.2B | Sep 2025 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
215
tok/s
129–344 · low confidence |
Pleias 1.0 1.2B ≈ | 1.2B | Dec 2024 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
215
tok/s
129–344 · low confidence |
Pleias-RAG-1B ≈ | 1.2B | Apr 2025 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
209
tok/s
178–251 |
Llama 3.2 1B | 1.2B | Sep 2024 | 2.2 GB | 131k tokens | Q8_0 | Comfortable |
|
207
tok/s
124–330 · low confidence |
MiniCPM-1.2B ≈ | 1.2B | Jun 2024 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
198
tok/s
119–317 · low confidence |
DeepSeek Coder 1.3B ≈ | 1.3B | Jan 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
198
tok/s
119–317 · low confidence |
DeepSeek-VL-1.3B ≈ | 1.3B | Mar 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
198
tok/s
119–317 · low confidence |
DigiRL ≈ | 1.3B | Jun 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
198
tok/s
119–317 · low confidence |
GLA Transformer 1.3B ≈ | 1.3B | Aug 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
198
tok/s
119–317 · low confidence |
Janus 1.3B ≈ | 1.3B | Oct 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
198
tok/s
119–317 · low confidence |
Kosmos-2.5 ≈ | 1.3B | Aug 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
198
tok/s
119–317 · low confidence |
Otter ≈ | 1.3B | May 2023 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
198
tok/s
119–317 · low confidence |
Phi-1 ≈ | 1.3B | Oct 2023 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
Speeds are estimates for a single request — one conversation at a time — calculated from memory bandwidth, model size and quantisation. Real throughput varies with the inference runtime and its version. Figures published by hardware vendors measure many simultaneous requests and are much higher.
On record
CMP 70HX full specification
Everything on record for this board, ordered by how much it bears on running a language model rather than by how a spec sheet would list it. Memory comes first because it decides the outcome; the rest is context.
Memory
The two specifications that decide what this card can run and how quickly. Capacity sets which models fit; bandwidth sets how many tokens per second they produce once they do.
- Memory size
- 8 GB
- Memory bandwidth
- 608 GB/s
- Memory type
- GDDR6X
- Memory bus width
- 256 bit
- Memory clock
- 1.19 GHz
The chip
Which processor is on the board and how it was manufactured. A smaller process size generally means more performance for the same power.
- Graphics processor
- GA104
- Architecture
- Ampere
- Generation
- Mining GPUs
- Foundry
- Samsung
- Process size
- 8 nm
- Transistors
- 17.4 billion
- Transistor density
- 44,400 K/mm²
- Die size
- 392 mm²
- Package
- BGA-2713
- Released
- 1 January 2021
Clock speeds
How fast the processor runs. Worth far less here than on a gaming benchmark: generating text is limited by memory bandwidth, so a higher clock barely moves the result.
- Base clock
- 1.37 GHz
- Boost clock
- 1.4 GHz
Processing units
What the chip contains. These drive graphics performance and matter mainly for processing a long prompt rather than for producing the answer.
- Shading units
- 3,840
- Texture mapping units
- 120
- Render output units
- 64
- Streaming multiprocessors
- 30
- Tensor cores
- 120
- Ray tracing cores
- 30
- L1 cache
- 128 KB
- L2 cache
- 4 MB
Theoretical performance
Peak arithmetic rates published for the board. These are ceilings that no real workload reaches, and generating text reaches a small fraction of them because it is limited by memory rather than arithmetic.
- Half precision (FP16)
- 10.7 TFLOPS
- Single precision (FP32)
- 10.7 TFLOPS
- Double precision (FP64)
- 167.4 GFLOPS
- Pixel rate
- 89 GPixel/s
- Texture rate
- 167 GTexel/s
The board
What it takes to physically install and power the card — the practical constraints that decide whether it fits the machine you already own.
- Suggested power supply
- 200 W
- Power connectors
- 1x 12-pin
- Bus interface
- PCIe 1.0 x4
- Slot width
- Dual-slot
- Dimensions
- 267 mm
Software support
Which graphics and compute interfaces the card supports. CUDA compute capability is the one that bears on inference: below 7.0 there are no tensor cores, and modern inference software falls back to slower code paths.
- CUDA compute capability
- 8.6
- DirectX
- 12.2
- OpenGL
- 4.6
- Vulkan
- 1.4
- OpenCL
- 3.0
- Shader model
- 6.8
Listings
Where to buy a CMP 70HX
No vendor is currently listing this card. Listings come from vendors who publish them here directly — browse the vendor directory to see who is selling what.
What the numbers mean
Memory: the specification that decides everything
Memory
8 GB
Bandwidth
608 GB/s
Largest model
Baichuan 1-13B
CMP 70HX carries only 8 GB of GDDR6X. That limits it to the smaller end of the catalogue, and a model has to fit entirely inside before it generates anything at all. A runtime actually gets about 7.2 GB.
Memory bandwidth reaches 608 GB/s across a bus of 256 bits. That is the number governing generation speed — arithmetic per byte read is small enough that the bus, not the cores, is what everything waits on.
That comes from a memory clock of 1.19 GHz. Widening the bus and raising the clock are the two levers a manufacturer has, which is why a card with unremarkable cores can still generate quickly.
Put together, the largest model that fits is Baichuan 1-13B, 13.3B, compressed to Q3_K_M and generating around 52.4 tokens per second.
The chip and how it was built
CMP 70HX is built on the graphics processor GA104, using the architecture Ampere from NVIDIA, as part of the generation Mining GPUs.
The chip is manufactured by Samsung, on a process of 8 nm, with a die measuring 392 mm², holding 17.4 billion transistors. A smaller process generally means more performance for the same power, though for language models it matters far less than the memory subsystem.
It was released in January 2021, roughly 5.5804549082089 years ago. Inference software support tends to follow hardware by a year or two, so a card of this age generally has mature, well-optimised code paths available to it.
Compute throughput, and why it matters less than it looks
FP16
10.7 TFLOPS
FP64
167.4 GFLOPS
Tensor cores
120
On paper CMP 70HX reaches 10.7 TFLOPS at half precision, and 10.7 TFLOPS at single precision. These are peak figures no real workload sustains, and generating text reaches only a small fraction of them — decoding is limited by memory rather than arithmetic, which is why a card can look enormously powerful here and still produce tokens at an ordinary rate.
Double-precision throughput reaches 167.4 GFLOPS. It has no bearing on running a language model — no inference runtime uses it — but it separates datacentre parts from consumer ones, since the latter deliberately restrict it.
The card carries 120 tensor cores across 30 streaming multiprocessors. These accelerate the matrix arithmetic at the heart of a transformer, and they are what make prompt processing — reading a long document before answering — dramatically faster than it would otherwise be.
Clocks run from a base of 1.37 GHz to a boost of 1.4 GHz. Worth far less here than on a gaming benchmark: raising the clock speeds up the arithmetic, and the arithmetic is not what generation is waiting on.
Cache and processing units
CMP 70HX has an L1 cache of 128 KB, backed by an L2 cache of 4 MB. Cache absorbs a share of the memory traffic that would otherwise hit the main bus, which is the one place on this page where a number other than bandwidth quietly affects generation speed — a large L2 lets more of the working set stay close to the cores.
There are 3,840 shading units, 120 texture mapping units, and 64 render output units. These drive graphics workloads and contribute to prompt processing, but they sit idle for much of the time a model spends generating a reply.
Power, size and installation
The board occupies dual-slot, measuring 267 mm long, and needs 1x 12-pin. Worth checking against the case and power supply already in the machine, since the largest cards need considerably more of both than a typical desktop provides.
It connects over PCIe 1.0 x4. The interface governs how quickly a model is loaded from disk into the card, not how fast it runs once there, so a narrower link costs a few seconds at startup and nothing thereafter.
The extremes
The largest AI models that run on a CMP 70HX
The biggest open-weight models that fit on this card, newest first. Each is shown at the best compression the card can hold.
The fastest AI models on a CMP 70HX
Where this card produces tokens quickest. Smaller models dominate here, because generating each token means reading the whole model out of memory once.
Step by step
How to work out the tokens per second of a CMP 70HX
You do not have to calculate anything by hand — the gputps.com calculator on this page has already worked it out for every model this card can hold. Reading off the answer takes six steps.
-
01
Start with the model, not the specification
The table lists 337 models this card runs. Search narrows the list by name or by size.
-
02
Set the context length you will actually use
Set the context to your real working length. Short questions cost almost nothing, but a long document can consume a large share of 8 GB it is often what pushes a large model over the edge.
-
03
Pin the comparison to one quality level
Each model is shown at the best compression this card can hold. A minimum quality hides the ones that only fit by being squeezed further than you would accept.
-
04
Take the range as the answer
Each speed is an estimate for a single conversation, with a range beneath it. The top end here is 258 tok/s on Gemma 3 QAT 1B. The same card and model vary by thirty to fifty per cent between inference runtimes.
-
05
Check the memory column before committing
Tight means it works today; comfortable means it still works when the conversation grows. Compare what each model needs against an available 8 GB.
-
06
Open the model to compare cards
Each model page repeats this calculation for the whole catalogue. Worth a look before deciding: it shows what else runs the same model, alongside CMP 70HX.
Answers
CMP 70HX — common questions
CMP 70HX— what is its memory bandwidth?
Memory bandwidth reaches 608 GB/s across a bus of 256 bits. This is the single best predictor of how fast it generates text, because producing each token means reading the entire model out of memory once.
CMP 70HX— what type of memory does it use?
It uses GDDR6X clocked at 1.19 GHz. HBM types are found on datacentre accelerators and carry far more bandwidth than the GDDR used on desktop cards, which is why they generate tokens considerably faster at the same capacity.
CMP 70HX— who makes it?
This is a product of NVIDIA, with the chip manufactured by Samsung, on a process of 8 nm.
CMP 70HX— when was it released?
It was released in January 2021.
CMP 70HX— how much cache does it have?
The L1 cache is 128 KB, and the L2 cache is 4 MB. Cache absorbs part of the memory traffic that would otherwise reach the main bus, so a larger L2 gives a modest lift to generation speed beyond what bandwidth alone predicts.
CMP 70HX— what are its TFLOPS?
It is rated at 10.7 TFLOPS at half precision and 10.7 TFLOPS at single precision. These are peak arithmetic ceilings rather than achievable rates, and text generation reaches only a small fraction of them because it is limited by memory bandwidth instead.
CMP 70HX— how many tensor cores does it have?
It has 120 tensor cores across 30 streaming multiprocessors. They accelerate the matrix arithmetic a transformer is built from, which mainly speeds up processing a long prompt rather than producing the reply.
CMP 70HX— does it support CUDA?
Yes. It reports CUDA compute capability 8.6. Capability 7.0 and above has tensor cores, which modern inference software uses; below that it falls back to slower code paths for quantised models.
CMP 70HX— what bus interface does it use?
It uses PCIe 1.0 x4. This governs how fast a model is loaded onto the card rather than how fast it runs once loaded, so it costs a few seconds at startup and nothing during generation.
CMP 70HX— is it good for running local AI models?
Its memory limits it to smaller models and its bandwidth gives usable, if unspectacular, generation speeds. In total it runs 337 of the models we track. Whether that is enough depends entirely on which model you want — the table above answers that directly.
CMP 70HX— can it run a model that does not fit in its memory?
Offloading past the card's 8 GB is possible and usually a false economy: the system-memory portion is slow enough to dominate the result.
Would two CMP 70HX cards be twice as fast?
Capacity adds, throughput does not. Two of them give you 16 GB of combined memory, at roughly the same generation speed as one.
CMP 70HX— which AI models can it run?
337 of the 679 open-weight language models we track fit on this card and can be run locally. The table on this page lists every one, with the memory it needs, the quantisation it runs at and an estimated generation speed.
CMP 70HX— what is the largest AI model it can run?
The largest model in our catalogue that fits is Baichuan 1-13B at 13.3B parameters, compressed to Q3_K_M. It generates roughly 52.4 tokens per second and needs about 7.2 GB of the card's memory.
CMP 70HX— how many tokens per second does it produce?
It depends on the model. The fastest model we track here is Gemma 3 QAT 1B at about 258 tokens per second, while larger models run proportionally slower because each token requires reading the whole model out of memory once. Speeds are estimates for a single conversation at a time.
CMP 70HX— can it run 7B models?
Yes. For example it runs MetaMath 7B (LLaMa finetune) at Q4_K_M, using about 6.5 GB of memory and generating around 85.0 tokens per second.
CMP 70HX— can it run 13B models?
Yes. For example it runs Gemma 4 12B at Q3_K_M, using about 6.5 GB of memory and generating around 58.2 tokens per second.
CMP 70HX— how much memory does it have?
This card has 8 GB of GDDR6X. Around a tenth is reserved by the inference runtime and the driver, leaving roughly 7.2 GB available for a model and its conversation.
The other direction
Looking at it from the other side?
This page starts from the hardware. If you already know which model you want and need to know what it takes to run it, start from the model instead.