"Support for Infinity Fabric Link GPU interconnect technology – With up to 84GB/s per direction low-latency peer-to-peer memory access1, the scalable GPU interconnect technology enables GPU-to-GPU communications up to 5X faster than PCIe® Gen 3 interconnect speeds2" source Radeon Pro Vega || Duo
if it was only pci 3.0 on mobile systems its still plenty quick but there is already IF2. it has reached its 2nd gen. just imagine the kind of power available in the apu only..for a mobile gpu+IF??
Is it IF2? and even then the die is not like chiplets guess what?? yup..smartshift.
@Ian: I may have missed it among the 380+ comments, but isn't this chip, especially the H line, also the dress rehearsal for the CPU part of the custom APUs that will power the PS5 and the next Xbox? Any rumors from AMD? The APUs must be in production already if SONY and MS want to ship the new consoles in mid/late 2020.
Gimme an 8C/16T desktop with integrated graphics, even if it's the lowest of the low, and my next build would use it. Just need basic graphics as a server, but the current desktop line is limited to 4C/8T with IGPU.
One possible error here. The mention of the 8 cores being broken into two 4 core CCX doesn't apply with Zen2, since AMD went to 6 and 8 core CCX units for the desktop. This means that for the laptop chips, we may be looking at a single 8 core CCX linked to the GPU instead of two four core CCX linked to the GPU.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
406 Comments
View All Comments
Aljon Pobar - Friday, January 10, 2020 - link
Competition is really good for Us consumers.carcakes - Friday, January 10, 2020 - link
"Support for Infinity Fabric Link GPU interconnect technology – With up to 84GB/s per direction low-latency peer-to-peer memory access1, the scalable GPU interconnect technology enables GPU-to-GPU communications up to 5X faster than PCIe® Gen 3 interconnect speeds2" source Radeon Pro Vega || Duohttps://www.notebookcheck.net/AMD-Radeon-Pro-Vega-...
&
https://www.amd.com/en/graphics/workstations-radeo...
if it was only pci 3.0 on mobile systems its still plenty quick but there is already IF2. it has reached its 2nd gen. just imagine the kind of power available in the apu only..for a mobile gpu+IF??
Is it IF2? and even then the die is not like chiplets guess what?? yup..smartshift.
carcakes - Friday, January 10, 2020 - link
it means that @LPDDR4x 4ch@2667 at least...and now with smartshift even more...how is memory access? which gen is it? is it all clocked the same?carcakes - Saturday, January 11, 2020 - link
How the discrete gpu with HT off as well as compared to smartshift?SharonTTurner - Monday, January 13, 2020 - link
..I dunno... ice lake is pretty competitive.hSharonTTurner - Monday, January 13, 2020 - link
nicepeevee - Monday, January 13, 2020 - link
WTVega?eastcoast_pete - Monday, January 13, 2020 - link
@Ian: I may have missed it among the 380+ comments, but isn't this chip, especially the H line, also the dress rehearsal for the CPU part of the custom APUs that will power the PS5 and the next Xbox? Any rumors from AMD? The APUs must be in production already if SONY and MS want to ship the new consoles in mid/late 2020.rrinker - Monday, January 13, 2020 - link
Gimme an 8C/16T desktop with integrated graphics, even if it's the lowest of the low, and my next build would use it. Just need basic graphics as a server, but the current desktop line is limited to 4C/8T with IGPU.Targon - Thursday, January 16, 2020 - link
One possible error here. The mention of the 8 cores being broken into two 4 core CCX doesn't apply with Zen2, since AMD went to 6 and 8 core CCX units for the desktop. This means that for the laptop chips, we may be looking at a single 8 core CCX linked to the GPU instead of two four core CCX linked to the GPU.