The rumours of Nvidia's upcoming Kepler line up of graphics cards are continuing to appear, this time with the GK104 being the subject of speculation.
Some journalists - yours truly included - posted a full rundown of specifications for the next generation graphics processing unit (GPU). Unfortunately the realisation that they might be fake came only after hitting the publish button, so forgive me if I take a slightly more dubious tack with these.
However, whether faked, rumoured, real, leaked or a combination of everything, today we have some new Kepler details. Take them with a pinch of salt, but here they are:
- 28nm production at TSMC
- Die size 340mm²
- 4 Graphics Processing Clusters (GPC)
- 4 Streaming Multiprocessors (SM) per GPC = 16
- SM 96 Stream Processors (SP) per SM = 1536 CUDA cores
- 8 Texture Units (TMU) per SM = 128 TMUs
- 32 Raster OPeration Units (ROPs) Chip clock (top model): 950 MHz
- 1250 MHz actual (5.00 GHz effective) memory, 160 GB/s memory bandwidth
- 256-bit DDR memory interface (up to GDDR5)
- 2048 MB (2 GB) memory amount standard
- 2.9 TFLOP/s single-precision floating point compute power
- 486 GFLOP/s double-precision floating point compute power
- Elimination of Hotclocks
No word from Nvidia on how legitimate these specifications are, and we likely won't get anything until the actual cards start hitting the review sites and their NDAs are up. Whatever these GPUs end up packing though, it better be impressive as Nvidia has already lost a lot of ground this generation, since AMD has beaten it to the punch by several months with its 7000 series.