NVIDIA’s Ampere GPUs Based on TSMC’s 7nm Node, Next-Gen Hopper GPUs Based on Samsung’s 5nm EUV Process, Alleges Rumor – Mass Production Begins This Quarter
In a new report, ChinaTimes has reported that NVIDIA's Ampere GPUs are indeed going to utilize TSMC's 7nm process node technology whereas the next-generation Hopper GPU family would be based on the Samsung 5nm EUV process node.
NVIDIA's Ampere Fabricated on TSMC's 7nm Process, Next-Gen Hopper GPUs on Samsung's 5nm EUV Node, Alleges Rumor
NVIDIA will be publishing a pre-recorded version of its GTC 2020 'Get Amped' keynote featuring Jensen Huang, the company's CEO, who will be presenting new innovations, announcements, and technologies. The core segment of the online keynote will focus on the Ampere GPU which will be unveiled along with technical information such as specifications and products that it would be featured inside.
According to ChinaTimes, NVIDIA has been confirmed as one of the biggest customers of TSMC's 7nm process node. This seems to indicate that NVIDIA's Ampere GPUs, at least for the HPC & the data center segment, will be based on TSMC's advanced chip manufacturing process. As soon as the news came out, Samsung released a statement in its Q1 earnings report, confirming that it will begin production of its 5nm (5LPE) EUV process in Q2 2020.
Samsung has already reportedly received orders for processors, graphics cards and various other products to utilize its 5nm process node. According to the report, most of these products are going to be utilized in consumer applications. Based on internal roadmaps, AMD is not expected to use Samsung foundries for its next-generation processors or graphics cards but it is very likely that the next-generation Hopper GPUs from NVIDIA will utilize Samsung's 5nm EUV node which would be replacing Ampere GPUs.
NVIDIA Tesla Graphics Cards Comparison
|Tesla Graphics Card Name||NVIDIA Tesla M2090||NVIDIA Tesla K40||NVIDIA Telsa K80||NVIDIA Tesla P100||NVIDIA Tesla V100||NVIDIA Tesla Next-Gen #1||NVIDIA Tesla Next-Gen #2||NVIDIA Tesla Next-Gen #3|
|GPU Name||GF110||GK110||GK210 x 2||GP100||GV100||GA100?||GA100?||GA100?|
|Transistor Count||3.00 Billion||7.08 Billion||7.08 Billion||15 Billion||21.1 Billion||TBD||TBD||TBD|
|CUDA Cores||512 CCs (16 CUs)||2880 CCs (15 CUs)||2496 CCs (13 CUs) x 2||3840 CCs||5120 CCs||6912 CCs||7552 CCs||7936 CCs|
|Core Clock||Up To 650 MHz||Up To 875 MHz||Up To 875 MHz||Up To 1480 MHz||Up To 1455 MHz||1.08 GHz (Preliminary)||1.11 GHz (Preliminary)||1.11 GHz (Preliminary)|
|FP32 Compute||1.33 TFLOPs||4.29 TFLOPs||8.74 TFLOPs||10.6 TFLOPs||15.0 TFLOPs||~15 TFLOPs (Preliminary)||~17 TFLOPs (Preliminary)||~18 TFLOPs (Preliminary)|
|FP64 Compute||0.66 TFLOPs||1.43 TFLOPs||2.91 TFLOPs||5.30 TFLOPs||7.50 TFLOPs||TBD||TBD||TBD|
|VRAM Size||6 GB||12 GB||12 GB x 2||16 GB||16 GB||48 GB||24 GB||32 GB|
|VRAM Bus||384-bit||384-bit||384-bit x 2||4096-bit||4096-bit||4096-bit?||3072-bit?||4096-bit?|
|VRAM Speed||3.7 GHz||6 GHz||5 GHz||737 MHz||878 MHz||1200 MHz||1200 MHz||1200 MHz|
|Memory Bandwidth||177.6 GB/s||288 GB/s||240 GB/s||720 GB/s||900 GB/s||1.2 TB/s?||1.2 TB/s?||1.2 TB/s?|
It will be definitely interesting to see an updated roadmap that clears this up during the GTC 2020 keynote. There also exists an opportunity for NVIDIA to go with Samsung's 5nm for its consumer-based graphics card lineup since mass production is indeed happening this quarter and GeForce cards replacing the Turing lineup could arrive by the end of 2020.
The NVIDIA GeForce RTX lineup was also unveiled in late 2018. We have internally confirmed with several sources that NVIDIA's AIB partners are aggressively clearing up their GeForce RTX (Turing GPU) inventories ahead of a big launch of a consumer-based GeForce lineup in Q3 2020. Q3 2020 also makes sense because that is where Computex 2020 has been shifted to but NVIDIA might select another venue to host its GPU (GeForce) announcement.
With that said, NVIDIA's Ampere GPUs are definitely going to shake things up in the HPC market with several variants already leaked and performance being rated at around 30 TFLOPs (FP32). We will keep you updated as more info comes prior to the 14th of May when NVIDIA will be presenting its next-gen GPU lineup.