- Mar 3, 2017
- 1,615
- 5,870
- 136
Ah you're right, I assumed because the OPN is the same it was one of the original batches, but looks like AMD doesn't change OPNs with newer siliconDoesn't the second line say this chip was produced in week nine of 2024 ?
That would be early march 2024.
FYI. The '+' means "AM5 and newer/future platforms, not an actual socket change by AMD.""AM5+ in 2026"AMD may have a new platform for upcoming Ryzen CPUs — AM5+ socket and Granite Ridge CPUs listed in a microcode extraction tool
AM5+ platform could come with new features.www.tomshardware.com
Why the heck would we be seeing a patch now then?
C'mon, HDMI is HDMI, huh ?My buddy just spent $2k on a gaming rig only to complain his game wasn't better.
Turns out he was 1920x1280 at 60fps because Windows defaulted to that with his HDMI. Does it really matter AMD/Intel when average consumers buy top end stuff only to not know the correct cables and settings matter?
Is he happier now? Doing 1440p?Turns out he was 1920x1280 at 60fps because Windows defaulted to that with his HDMI.
I think Paul mistook ballpoint for ballpark 😅
People like you keep giving fantasy sources more potential hit. If you know and suspect these are made up, there's nothing to discuss other than inviting others to waste more time on the noises. For the record, I didn't even look at the words in your screenshot.Some info from RTG
What true or no true. We don't know
We can still discuss about it. Maybe Kepler, know something, what real/not.People like you keep giving fantasy sources more potential hit
It's true, though it is Turin @ 500W vs Genoa @ 400W.Some info from RTG
What true or no true. We don't know
AI PC meme gimmick is for laptops, includes rebates too.Or perhaps not comply with the requirements by Microsoft at all, so the Zen5 DT based PCs will be placed in the "non-AI PC" category in stores?
no.but it'll probably be a check box feature that is going to determine a lot of people's buying decisions in the near future.
That still good, if 1.5though it is Turin @ 500W vs Genoa @ 400W.
Sorry I have to put a little bit of a dampener on things. Now obviously I can't confirm 100% the SPECInt numbers are true (I mean, everyone I've talked to has tried to convince me they are, but like... come on those numbers are crazy, can you blame me?) but even with all the craziness I have to point out that the Genoa numbers are supposedly 400W, the Turin numbers are at 500W. They're not quite as apples-to-apples as the slide makes them seem.Some info from RTG
What true or no true. We don't know
1.25X, the actual core power is lower for reasons you can guess.perf boost at ~1.3x the power
If that s 1.5x the perf at 1.25x the power then it amount to 10% lower perf at same 400W power, that is 1.35x the perf at same power, not an extraodrinary number all things being equal given that it s a server plateform, it would be much more impressive for a DT part.Sorry I have to put a little bit of a dampener on things. Now obviously I can't confirm 100% the SPECInt numbers are true (I mean, everyone I've talked to has tried to convince me they are, but like... come on those numbers are crazy, can you blame me?) but even with all the craziness I have to point out that the Genoa numbers are supposedly 400W, the Turin numbers are at 500W. They're not quite as apples-to-apples as the slide makes them seem.
When you factor in IOD power on server and look at just CPU core performance, you're actually really looking at a 1.5x perf boost at ~1.3x the power, around abouts. Still incredibly good for a large core on the same node, but not as crazy as what's flying around sounds like. I really feel like I have to throw a but of water on the fire here.
Again, even assuming the numbers are true
This about the same 36% uplift we got between 3950x and 5950x single thread in this benchmark. Harder to judge multi. 3950x to 5950x was only about 13% in this benchmark. 5950x to 7950x was 46% but the 7950X also has a higher power limit. 7950X3D is 42% faster than 5950X with a similar power limit but the vcache helps a bit in this benchmark. Seems plausible. In that context it doesn't seem all that earth shattering though.For reference, rough numbers for Zen 4:
7950X Single ~125, Multi ~2200
7900X Single ~120, Multi ~1650
~36% improvement if real. Don't know if that's plausible.
Cinememe is the least interesting thing ever.Not that i trust RTG, but If by chance those numbers turn out real, i will consider buying for sure. Not for 1000 though, lol.
I don't think Kepler & Adroc are lying.For reference, rough numbers for Zen 4:
7950X Single ~125, Multi ~2200
7900X Single ~120, Multi ~1650
~36% improvement if real. Don't know if that's plausible.
What are these AVX512 heavy apps? Aside of PS3 emulator.I don't think Kepler & Adroc are lying.
How well will that 40%+ perf translate into apps is a different story, it may be stellar in some, less so in others.
The other, unspoken aspect, is AVX512 support. Zen4 has a trimmed AVX512 support, where they run some kind of dual 256bit design, I forget the details, but it's more of a "can do AVX512" than a "is custom made for AVX512".
The likelihood of AVX512 heavy apps being massively faster, even way beyond 40%, is very high.
Some math stuff but they're all limited by the bandwidth available.What are these AVX512 heavy apps? Aside of PS3 emulator.