You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
v3.3: Add latest CPUs and GPUs. Add Cloud Hypervisor.
We have Zen-6 on the Horizon and we might see Oryonv3 (Qualcomm) based
server CPUs.
Add nVidia Blackwell (b), AMD CDNA 4, RDNA 4 and 5, intel BattleImage (4).
I was wrong with my assumption that CDNA and RDNA are very similar, so
use .1 versions for RDNA. This means that server/supercomptuing GPUs
don't change their names. (Consumer GPUs would, but those are unlikely
to be used in clouds -- I'm not aware of any.)
Add Cloud Hypervisor (intel / cyberus) as new Hypervisor (chy).
Signed-off-by: Kurt Garloff <kurt@garloff.de>
It is recommended to leave out the `0` when specifying the old generation; this will
380
381
help the parser tool, which assumes 0 for an unspecified value and does leave it
@@ -386,7 +387,7 @@ out when generating the name for comparison. In other words: 0 has a meaning of
386
387
We don't differentiate between Zen-4 (Genoa) and Zen-4c (Bergamo); L3 cache per
387
388
Siena core is smaller on Bergamo and the frequency lower but the cores are otherwise
388
389
identical. As we already have a qualifier `h` that allows to specify higher frequencies
389
-
(which Genoa thus may use more and Bergamo not), we have enough distinction
390
+
(which Genoa thus may use and Bergamo not), we have enough distinction
390
391
capabilities. The same applies to Zen-5 (Turin) and Zen-5c (Turin Dense).
391
392
For intel with the server E-cores (Crestmont), these received their own
392
393
generation assignment, as the difference to the server P-cores (Redwood Cove)
@@ -437,9 +438,9 @@ Note that the vendor letter X is mandatory, generation and processing units are
437
438
|`A`| AMD | compute units (CUs) |
438
439
|`I`| Intel | execution units (EUs) |
439
440
440
-
For nVidia, the generation N can be f=Fermi, k=Kepler, m=Maxwell, p=Pascal, v=Volta, t=turing, a=Ampere, l=Ada Lovelace, g=Grace Hopper, ...,
441
-
for AMD GCN-x=0.x, RDNA1=1, C/RDNA2=2, C/RDNA3=3, C/RDNA3.5=3.5, C/RDNA4=4, ...
442
-
for Intel Gen9=0.9, Xe(12.1/DG1)=1, Xe(12.2)=2, Arc(12.7/DG2)=3 ...
441
+
For nVidia, the generation N can be f=Fermi, k=Kepler, m=Maxwell, p=Pascal, v=Volta, t=turing, a=Ampere, l=Ada Lovelace, g=Grace Hopper, b=Blackwell, ...,
442
+
for AMD GCN-x=0.x, CDNA-x=x, RDNA-x=x.1, RDNA-3.5=3.5, UDNA-x=x
443
+
for Intel Gen9=0.9, Xe(12.1/DG1)=1, Xe(12.2)=2, Arc(12.7/DG2)=3, BattleImage(20.0)=4, ...
443
444
(Note: This may need further work to properly reflect what's out there.)
444
445
445
446
The optional `h` suffix to the compute unit count indicates high-frequency GPU compute units.
0 commit comments