Saturday, October 24th 2026

AMD EPYC 9996 Geekbench Listing Offers First Look at Zen 6 Single-Core Gains

AMD's Zen 6 just got one of its first rough per-clock comparisons against Zen 5, and the early numbers look good. Leaker HXL spotted a Geekbench 7.1.0 listing for the 256-core EPYC 9996, the flagship "Venice" chip AMD launched in July, scoring 2,847 points in single-core on a Dell PowerEdge R9815 server running Linux. User three twenty six on X then locked a Ryzen 7 9700X to 4.1 GHz, matching the EPYC 9996's maximum boost, and got 2,444 points on the same Geekbench version. That works out to roughly 16.5% higher single-core performance for Zen 6 at the same frequency. Apparently, Zen 6 came out ahead in all 15 single-core subtests, with gains ranging from 8.2% in File Compression to 25.7% in HTML5 Browser.

Of course, these scores only give us a general idea about the gains. The EPYC 9996 uses the dense Zen 6c cores, not the full Zen 6 cores expected in desktop Ryzen chips, so it's hard to say how closely these numbers will carry over. The two systems are also wildly different, with a server platform on one side and a 9700X running DDR5-6000 CL34 under CachyOS Linux on the other, and the EPYC listing has since been pulled from the Geekbench database. Still, a double-digit per-clock gain from the dense core variant is quite encouraging. AMD has already confirmed that "Olympic Ridge" desktop CPUs will stay on AM5, though they reportedly won't arrive until 2027, so we can expect more comprehensive performance leaks well before then.
Sources: Geekbench 7.1.0, HXL on X, three twenty six on X, via Wccftech
Add your own comment

19 Comments on AMD EPYC 9996 Geekbench Listing Offers First Look at Zen 6 Single-Core Gains

#1
Makaveli
This is looking very good considering this is a Zen 6c Core and not the full core we will see on the desktop with higher clocks.

In this benchmark that Zen 6c core is only alittle slower than my Zen 5 X3D core

6.4% slower at 4.1 vs 5.4 Ghz

Posted on Reply
#2
matar
With 256 cores 512 threads on one chip no need for a GPU run the game on CPU :laugh:
Posted on Reply
#3
user556
Bloody AI shit, messing everything up.
Posted on Reply
#4
tfp
1 GB of L3 makes it hard to call this a core to core comparison.
Posted on Reply
#5
jdrch
Makaveliwe will see on the desktop with higher clocks.
I highly doubt we see anything before 2030. And that's if desktop computing hasn't pivoted completely to UMA by then.
Posted on Reply
#6
810Cat
user556Bloody AI shit, messing everything up.
What does AI have to do with the article???
Posted on Reply
#7
RootinTootinPootin
MakaveliThis is looking very good considering this is a Zen 6c Core and not the full core we will see on the desktop with higher clocks.

In this benchmark that Zen 6c core is only alittle slower than my Zen 5 X3D core

6.4% slower at 4.1 vs 5.4 Ghz

that still doesn't translate to actual performance when it comes out, we've all seen this kind of teaser before from them, this will go again as a wild hype train for fanboi's..
Posted on Reply
#8
qcmadness
MakaveliThis is looking very good considering this is a Zen 6c Core and not the full core we will see on the desktop with higher clocks.

In this benchmark that Zen 6c core is only alittle slower than my Zen 5 X3D core

6.4% slower at 4.1 vs 5.4 Ghz

Extra L3 cache would not make a difference in "throughput".
But a 10%+ gain in IPC is always welcomed.
tfp1 GB of L3 makes it hard to call this a core to core comparison.
Except that the L3 cache slices are not shared between CCDs.
Posted on Reply
#9
melkor_unlimited
qcmadness
tfp1 GB of L3 makes it hard to call this a core to core comparison.
Extra L3 cache would not make a difference in "throughput".
But a 10%+ gain in IPC is always welcomed.

Except that the L3 cache slices are not shared between CCDs.
1024MB of L3 is divided among 8 chiplets (32 Zen 6c per one).
That means 128MB of L3 with direct access per chiplet.
7800x3D/9800x3D/9850x3D have 96MB per chiplet.
7700x/9700x got 32MB per chiplet.

128MB L3 may help a "little" in ST task but its not 32 times bigger.
Posted on Reply
#10
Chrispy_
This is too different from the AM5 chips to really pull any meaningful conclusions by comparing against a 9700X

What you really need to do is compare a single-threaded geekbench score against the same test on a Zen5 Epyc 9965, then scale the score for clockspeed (4.1 vs 3.7GHz)
Posted on Reply
#11
xxxZAURxxx
AMD is moving forward like a train with failed brakes.
Posted on Reply
#12
csendesmark
Can't wait for the new top Zen6,Especially if the leaks were correct and will have 24 cores!
Posted on Reply
#13
Makaveli
RootinTootinPootinthat still doesn't translate to actual performance when it comes out, we've all seen this kind of teaser before from them, this will go again as a wild hype train for fanboi's..
I'm aware this is just Geekbench. We are not going to see real actual performance numbers until release.
Posted on Reply
#14
user556
810CatWhat does AI have to do with the article???
Zen 6 was launched in July and we still don't have any. They're all going into servers only.
Chrispy_This is too different from the AM5 chips to really pull any meaningful conclusions by comparing against a 9700X

What you really need to do is compare a single-threaded geekbench score against the same test on a Zen5 Epyc 9965, then scale the score for clockspeed (4.1 vs 3.7GHz)
Like this: browser.geekbench.com/v7/cpu/47989

Epyc 9965 = 1838 points at 3.7 GHz.
Epyc 9996 = 2847 points at 4.1 GHz.

1838 x 4.1 / 3.7 = 2037. 2847 / 2037 = 1.3976.
So 40% faster. Even further apart than the Ryzen 9700X result.

browser.geekbench.com/v7/cpu/157801
2164 x 4.1 / 3.7 = 2398 points, => 2847 / 2398 = 1.1872. Almost 19% faster against a cherry picked Epyc 9845 (Zen 5C).
Posted on Reply
#15
csendesmark
user556Zen 6 was launched in July and we still don't have any. They're all going into servers only.
For now, Olympic Ridge coming to desktop in 2027-H1. (Maybe CES?)
Posted on Reply
#16
Makaveli
I think CES 2027 should be announcement.

Then retail launch in 2-3 months?
Posted on Reply
#17
Tess
DrUsmanSaleemStill, a double-digit per-clock gain from the dense core variant is quite encouraging.
It is? No, I don't think so. That's the standard improvement with every new design. Intel was doing that for years and AMD took to it ever since ZEN came out. A 20~25% increase in some benchmarks, 10~15% is others and a few remain about the same. This is standard progression, nothing special or encouraging other than to say it's not a dud or a step backwards.

Those marks are actually discouraging to me. Only the standard typical bump? Sh!t! I was expecting a lot more from Zen6! The claims about the near elimination of 1% and 0.1% lows in gaming will likely not be realized as hyped and advertised if all we're getting is the standard bump. Insiders have said Zen6 will be the biggest thing since Zen1 was released. This leak makes that sound like total BS!

Dang it, I went for a DDR5 system for nothing? Damn! :D

Ofc, there's always hope this "leak" isn't representative of Zen6 and it actually is encouragingly impressive - <crosses fingers>, <knocks on wood>, <buys more crystals>
Posted on Reply
#18
user556
Tesselator... The claims about the near elimination of 1% and 0.1% lows in gaming ...
That's still to be discovered. Nothing here says either way.

On the other hand, 0.1% lows are more about rendering engine behaviour, and game devs utilising it effectively.
Posted on Reply
#19
Tess
user556Nothing here says either way.
"Insider" scoops have claimed so, even if it's not specifically here in this thread.
Posted on Reply
Add your own comment
Nov 12th, 2026 04:39 CST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts