Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)

[DisEnchantment]

Speculate at will

 

[jdubs03]

Geekerwan managed to get spec results, +8.6% int & 26% fp.

View attachment 104679

I was looking at this too. Since its in this thread I wanted to post the further comparison of its competition. And ask a question:
Anyone have ideas why the i9-14900K values are the same, but the M4 and M3 FP values (but not INT) values are different?

I was thinking it could be due to the flags/compilers chosen, but the i9 would be different too then. The clock speeds are the same too.

 

[HurleyBird]

I don't cherry pick reviews looking for results I want. HWUB is my goto. Especially not interested in some esoteric Linux results. Newsflash: Linux is not "EVERTHTHING else".

Ever hear the term "regression to the mean"?

Phoronix always runs a huge number and variety of tests. In this case, something like 400 of them. It's the exact opposite of cherry picking. They found a 15.9% uplift over the 7700X, and with so many data points, that's probably about where it will stay even if you 10x the number of tests. Even if you took every other review site and averaged their results, there's probably less coverage than with Phoronix due to the large overlap in test suites.

 

[inf64]

Computerbase has 5.5Ghz Zen 5 at 14% better than 7950x @ 5.7Ghz. At least in their ST mix (new suite), Zen 5 shows >16% IPC increase.

 

[Hitman928]

Why? Most likely nearly 2025 for ARL, which probably will at best (for Intel) trade blows with Zen 5. Most of 2026 will be ARL refresh, which probably wont bring much to the table. Maybe 2027 will bring Nova Lake, which is rumored to be a significant step up, but the way Intel is executing now, I would not count on it either being on time or delivering the expected performance uplift. So if Zen 6 is late 2026 or early 2027, they should be fine.

I am talking on desktop. Mobile, I have no idea. I am not interested in that area, and the line-ups from both manufacturers are seriously confusing. So I dont know how things will stack up there, although AMD has a serious power advantage now, so even if Intel makes big gains in efficiency, I think AMD should also be fine in mobile.

Obviously the future is not certain, but 2.5 - 3 years is a long time between architectures and gives a very wide opening for competition to overtake your products. This is especially true on mobile and server where it's not only Intel that AMD has to worry about. In that time frame, Intel will probably have 2 generations out, beyond ARL. Of course, Intel could implode as a company by then, but they may also turn things around successfully. Additionally, ARM cores seem to be iterating pretty quickly as well, though I'm not as familiar with their roadmaps. If I were AMD, that wouldn't leave me feeling confident with my place in the competitive landscape come 2026.

 

[igor_kavinski]

It's very fast in branchy web stuff, but somehow not in video games?

Someone needs to test with older games. Modern games stream a lot of data in and out of RAM and since there isn't much improvement on that front, Zen 5 might be extremely memory bound in games. One could test, for example, how going from single channel to dual channel affects a game's fps on both 7700X and 9700X. If the 9700X sees a bigger improvement with dual channel, then it clearly craves more membw and that would bode well for the gaming performance of the 9800X3D.

 

[Hitman928]

Computerbase has 5.5Ghz Zen 5 at 14% better than 7950x @ 5.7Ghz. At least in their ST mix (new suite), Zen 5 shows >16% IPC increase.

View attachment 104702

It's a mix of INT and FP tests. If you take SPEC 1T results you get basically the same IPC increase if you average the INT and FP results together.

 

[biostud]

Lets just say there is more left in the tank
His latency numbers seem pretty medicore

But i agree that you can get much more out of Zen5 with a overclock, compared to Zen4.
These cpus are powerlimited at default settings
View attachment 104682

Yeah they should maybe just have named them 9700 and 9600 without the X.

Is there any chance that the redesigned core will benefit more from extra cache, so that it to a higher degree will help in memory bandwidth limited situations?

 

[techjunkie123]

My final verdict after reading a lot of reviews:

1. gaming is barely better than Zen 4.
2. Server/DC/content performance and efficiency looks great.

I'm looking forward to Turin!

 

[inf64]

It's a mix of INT and FP tests. If you take SPEC 1T results you get basically the same IPC increase if you average the INT and FP results together.

Well the real world workloads are also mostly a mix of both worlds, so no surprise there.

 

[poke01]

I was looking at this too. Since its in this thread I wanted to post the further comparison of its competition. And ask a question:
Anyone have ideas why the i9-14900K values are the same, but the M4 and M3 FP values (but not INT) values are different?

I was thinking it could be due to the flags/compilers chosen, but the i9 would be different too then. The clock speeds are the same too.

View attachment 104701

Probably cooling, the M4 is an a 5.1mm tablet. I doubt it maintained its frequency.

 

[HurleyBird]

Zen6 is 2027

Be quiet now failed leaker.

Going straight from unrealistic hype train conductor to doomer FUD propaganda...

It's of course possible that Zen 6 development will have problems and therefore arrive in 2027.

But it's not really possible that AMD is planning to release Zen 6 in 2027.

 

[gdansk]

Be quiet now failed leaker.

It's of course possible that Zen 6 will development will have problems and thus arrive in 2027.

But it's not really possible that AMD is planning to release Zen 6 in 2027.

If it takes until 2027 to get a 10% improvement over Zen 5 then I wager they'll lose every single market to bog-standard licensable commodity ARM cores.

 

[poke01]

There is no way Zen6 is 2027. It’s going to be 2026 at worst late 2026.

 

[jdubs03]

Probably cooling, the M4 is an a 5.1mm tablet. I doubt it maintained its frequency.

I see where the difference in the M3 is actually from the 2nd image (had to look at the video on the HX 370).

For the M4 yeah, I went back and he used LN2. That’ll do it lol. So for a passive cooled system that’s quite good. Will be interesting to see what the MacBooks can score.

From my interpretation, I think the pessimism is a bit overdone for the 9700X. IPC improvement is pretty decent. Sure it’s not a gaming workhorse but it’s also not an X3D. So could be better could be worse.

 

[igor_kavinski]

But it's not really possible that AMD is planning to release Zen 6 in 2027.

Well, if it needs a mobo change due to DDR6, 2027 might be right. There might be a Zen 5.5 in 2026 that has some features of Zen 6 but uses DDR5.

If we assume that Intel jumps on the DDR6 bandwagon in 2026, AMD may wait one year for the DDR6 prices to come down.

 

[DisEnchantment]

There is no way Zen6 is 2027. It’s going to be 2026 at worst late 2026.

If they take that much to time to get to Zen 6, AMD might as well tape out a stock ARM design and ramp down their x86 IP development, might save some big bucks

 

[Joe NYC]

Italian guy was better, he got at least few days of fame, this one will have only an hour

Daylight saving time?

 

[CakeMonster]

Well, if it needs a mobo change due to DDR6, 2027 might be right. There might be a Zen 5.5 in 2026 that has some features of Zen 6 but uses DDR5.

If we assume that Intel jumps on the DDR6 bandwagon in 2026, AMD may wait one year for the DDR6 prices to come down.

Sounds very unlikely to me with DDR6 so soon, that's why I'd hope AMD has a plan for DDR5 platform with new IO die and 3N in 2026.

 

[blackangus]

When you have to cherry pick Tom's, you know you are grasping at straws.

Most of the reputable sites are showing about 5% difference in both Windows games and productivity.

First Im not cherry picking anything, thats a (bad) assumption.
I literally said - "Thats the only thing I have had time to look at" and you still feel the need to insult.
Second 5% more performance for much less power consumption still supports my original statement of more performance for less power.

 

[In2Photos]

It's almost like AMD has decided the non-3D is for productivity and the 3D is for gaming.....

I think most would be ok with this if they add a lower end X3D, like a 9600X3D or 9700X3D, ala 5000 series.

 

[igor_kavinski]

I think most would be ok with this if they add a lower end X3D, like a 9600X3D or 9700X3D, ala 5000 series.

I would prefer it if they called it something like 963D or 96X3D. The 00 is superfluous for these single CCD parts since they never released any xx50 single CCD part.

 

[Rheingold]

It'd matter if it provided tangible benefit in performance, but most people will just open their faviourite tech-youtuber (GN or HU),

Those two are not the primary sources in Germany. The swing towards higher efficiency and lower temperatures is reflected much more positive over here where electricity costs are much more relevant for buyers. The fix for all those "too hot" posts was usually "use eco mode".

 

[Rheingold]

Crazy lower numbers, from LTT Youtube test.

Yep, even lower there. I suppose LTT was using an AIO and not an air cooler.

 

[Jan Olšan]

Hmm, regrettable (and unexpected) that the gaming performance isn't improved much but I have to say I don't quite care about that. I wonder if there will be some AGESA/etc improvements in gaming. There is certainly better chance of gaming improving over time than application software improving. No idea, did post-launch updates to firmware/AGESA improve performance on Zen 4 (non-X3D)?

That said, the non-gaming performance is solid. I went through the HWCooling tests, lots of (not avx512-related) 10%+ IPC gains in single thread tasks. And it's stuff like Geekbench 5, the browser benchmarks.

IMHO this core will withstand the test of time, despite the hysteria and dud/fail clickbaiting in some places

 

[Kepler_L2]

But it's not really possible that AMD is planning to release Zen 6 in 2027.

But they are