AMD Ryzen 7 9800X3D Linux Performance: Zen 5 With 3D V-Cache

[geerge]

Thanks for the review.
I have mixed feeling about this chip. As skeevy420 pointed out this chip uses substantially more power than 7800X3D, so I am not sure how fair is is to say that it "delivered 1.34x the performance of 7800X3D". Also "Compared to the Ryzen 7 9700X, the Ryzen 7 9800X3D offered 9% better performance." This also is not fair comparison, it should be compared against the 9700X at cTDP 105W. Here the the gain is just shy of 4%.
Yes, there are a handful of productivity tasks where the extra cache helps (not just games), however, they are not that many and the 9700X3D is substantially more expensive than 9700X (and the 12 core 9900X too!).

I really appreciate your in depth reviews, if I have one suggestion perhaps it would be to split the geomean in two parts: one for single core tasks and one for multi-core tasks. I think the many single core tasks skew the geomean in favor of CPUs with lower cores. Another issue may be tasks where the execution time is dominated by the memory bandwidth more than the CPU speed.

It uses more power because the 7800x3d could not because the cache was on top hindering thermals, which has been corrected this gen. So the 7800x3d ran at a better point on the efficiency curve because it had no choice, which IMO is where I like to run things anyway. If you run the 9800x3d in efficiency mode you'll probably find a similar performance bump from 7000 to 9000 that the non-x3d parts have.

[pWe00Iri3e7Z9lHOX2Qx]

Pretty happy with the $391 I paid for my 7900X3D earlier this year . Also, I'll repeat my cautionary advice to carefully check the specs of any X870/E boards you are shopping for. Adding USB4, and on many boards cranking up the number of PCIe 5 NVMe drives, means you often end up with worse tradeoffs for shared bandwidth than we saw with X670/E. One of the few times I'd be more interested in previous gen motherboards.

[skeevy420]

It uses more power because the 7800x3d could not because the cache was on top hindering thermals, which has been corrected this gen. So the 7800x3d ran at a better point on the efficiency curve because it had no choice, which IMO is where I like to run things anyway.

I have one so I know from first hand experience that the thermals aren't necessarily the case. Mine gets thermal throttled at 90c under Prime95 without the tweaking and it's 84C with the tweaks. Even with the cache on top there's still a bit of thermal headroom assuming the person is using a decent CPU cooler. Ignoring GPU junction temps, I don't think I've ever had a game make my system run hotter than 75C with or without tweaking.

There's also the fact that the better 7000X3D chips have their X3D cores set to run at higher clocks. That right there proves that there's no technical reason the 7800X3D has to be the lowest clocked 7000X3D chip. It's obviously artificially limited to make the other 7000X3D products look better. Even with those limitations it's still one of the best CPUs to have.

I wish AMD would release a firmware update to increase the 7800X3D core speeds to that of the 7950X3D 3D cores. It's not like we have the other 8 non-X3D cores to keep cool, too...

If you run the 9800x3d in efficiency mode you'll probably find a similar performance bump from 7000 to 9000 that the non-x3d parts have.

Those are the numbers I'm interested in seeing. 10% more performance with 40% more power usage is expected based on the AMD FX and Intel Lake histories. It's also not very exciting. Performance with the same power usage tells a more exciting story between the CPUs.

[rodja]

It uses more power because the 7800x3d could not because the cache was on top hindering thermals, which has been corrected this gen. So the 7800x3d ran at a better point on the efficiency curve because it had no choice, which IMO is where I like to run things anyway. If you run the 9800x3d in efficiency mode you'll probably find a similar performance bump from 7000 to 9000 that the non-x3d parts have.

I agree with you. I think if you run the 9800X3d at 65W, it'll be 4% better than 7800X3D.

[Ezakimak]

Really needs the 9800 included to be a complete comparison so we can see the difference clock-for-clock with and without the new 3D V-cache design...

[rodja]

Really needs the 9800 included to be a complete comparison so we can see the difference clock-for-clock with and without the new 3D V-cache design...

It is included. it's the 9700X both at stock and at 105 cTDP.

[Michael]

Really needs the 9800 included to be a complete comparison so we can see the difference clock-for-clock with and without the new 3D V-cache design...

There isn't any released Ryzen 7 9800X (non-X3D) design yet so that's why it wasn't included....

[Ezakimak]

There isn't any released Ryzen 7 9800X (non-X3D) design yet so that's why it wasn't included....

Maybe run the 9800X3D once through with cores disabled and frequencies set to match the 9700X ? Wouldn't be perfect but could get us a better idea.

[Squisher]

It may need some more testing but it kind of shows how zen5 could perform if amd did their due diligence and also updated their infinity fabric and io die. Guess they wanted to go cheap with zen5 to hope that zen6 really hits hard. But the air is getting thinner and thinner for x86 designs. Like from intel everything surrounding the cpu is still pack leading

[JEBjames]

Michael

Great review.

Minor typo Page 1

"while total it provides 104MB of cache." should be "while in total"