News Ryzen 7 9800X3D trails Core i9-14900K in leaked PugetBench benchmarks — upcoming Zen 5 3D V-Cache chip shows good performance uplift over Ryzen 9000

[Admin]

AMD's Ryzen 7 9800X3D was benchmarked in PugetBench and performed excellently against the Ryzen 7 9700X and even the Ryzen 9 7950X3D.

Ryzen 7 9800X3D trails Core i9-14900K in leaked PugetBench benchmarks — upcoming Zen 5 3D V-Cache chip shows good performance uplift over Ryzen 9000 : Read more

 

[bit_user]

Puget makes clear that some of their video-related benchmarks benefit heavily from Intel's QuickSync hardware codec acceleration. Given that AMD is still using the same IO Die as in Zen 4, it was obvious they weren't going to improve on that front.

Also, it should always be noted that Puget's CEO sits on Intel's Advisory Board. So, he, his company, and their benchmarks should not be treated as an unbiased source of truth.

The article has some merit, in analyzing the gen-on-gen performance between the n800X3D models and comparing the 9700X to 9800X3D. However, due to Intel's QuickSync video advantage, it shouldn't be treated as basis for comparing pure compute performance between AMD and Intel.

 

[Elusive Ruse]

Puget is a mouthpiece for Intel, that said, I am wondering why they are playing with an X3D to begin with? Their scope of business and their clients are not in line with the intended target market of the 9800X3D.

 

[bit_user]

I am wondering why they are playing with an X3D to begin with? Their scope of business and their clients are not in line with the intended target market of the 9800X3D.

I had the same question, but the article starts out saying that it was an independent actor who simply ran Puget's benchmark suite (which you can presumably download and run for yourself).

It's also a bit rich that it's comparing against the best entries in the scores database, which includes machines of unknown memory configuration and overclocking status.

 

[DingusDog]

Hmmm cherry picked benchmarks against a 14900k instead of the core ultra 285k I wonder why...

 

[Elusive Ruse]

I had the same question, but the article starts out saying that it was an independent actor who simply ran Puget's benchmark suite (which you can presumably download and run for yourself).

It's also a bit rich that it's comparing against the best entries in the scores database, which includes machines of unknown memory configuration and overclocking status.

Ah that’s my bad then, but seeing 9800X3D beating 9700X comprehensively, the different configurations being the caveat; tells me the X3D has got some ST chops.

 

[jeremyj_83]

A 24c/32t 2023 top of the line CPU being faster than an 8c/16t 2024 mid range CPU isn't shocking at all.

 

[aberkae]

My predictions is the 9800X3D will sell out instantly. Bummer that Ryzen R9 x3d parts are launching later. I wonder how many Intel loyalists are going to jump ship after all these years.

 

[Thunder64]

Surprised Paul Alcorn didn't write this. SMDH.

 

[Joseph_138]

It doesn't matter if it's slower. The only thing that's going to matter is that it isn't burning up motherboards.

 

[TeamRed2024]

I wonder how many Intel loyalists are going to jump ship after all these years.

Already done... and it wasn't a difficult decision at all.

 

[TheHerald]

The new latest and greatest unreleased r7 loses to last year's much cheaper i7. Comments

"he is on intels board"
"booho it has less cores"

As if that's an excuse. Amd will release a single core chip for 599 and people will defend its performance cause it has only one core. As if that is not exactly the issue, that it only has one core. Crazy times we live in

 

[YSCCC]

Already done... and it wasn't a difficult decision at all.

For production workload, stability is always the main thing (you can’t transcode to halfway and restart if it crashes 50min later), so beside marks, instability is no.1.

For gaming, cache benefits a ton.

So conclusion is it isn’t a difficult choice, if my current 14900k fries I likely won’t even bother rma it, kept as a screwed memorial and move on to a new platform

 

[TheHerald]

Cache benefits a ton in light scenes (looking at the sky) when the data fits into the cache. Harder scenes that don't fit the cache and it hits a brick wall.

Factorio (and MSFS) are prime examples of that, since that is the game that benefits with cache the most. You bench the small 10k maps and holy cow, the 7800x 3d is over twice as fast as a 14900k. Over TWICE. That's insane. Problem is, they are both spitting fps into the hundreds, around 300-350 fps for the 14900k, and around 600-700 for the 7800x 3d.

Then you move into the actual big complicated maps of factorio and.....the x3d is even losing to alderlake and this time it actually matters cause it barely stays above 60. So yeah, what's the point of having a lead in light scenes when every cpu can spit lots of fps and then completely crumble in the harder scenes when you actually need the fps.

Especially since the game is basically restricted to 60ups, the goal on this one is to stay above 60 with as big a map as possible. The 3d chips fall flat on their face here. But then you watch hwunboxeds review and you have to pretend that they are 100% faster than the competition in this game, lol

https://factoriobox.1au.us/results?...9f36f221ca8ef1c45ba72be8620b&vl=&vh=&sort=ups

Same happens in eg. Cyberpunk and MSFS. Digital foundry made a video with live footage comparing the cpus in the actual hard scenes of those games and yeah, results were nowhere near close to what you'd expect. Reviews have the 7800x 3d hitting over 200 fps in cyberpunk, actual footage from a hard area (tom's dinner) and it drops to low 60s.... Literally 3 fps higher than the 9950x, a CPU that is unsuitable for gaming according to the experts here

 

[stuff and nonesense]

Cache benefits a ton in light scenes (looking at the sky) when the data fits into the cache. Harder scenes that don't fit the cache and it hits a brick wall.

Factorio (and MSFS) are prime examples of that, since that is the game that benefits with cache the most. You bench the small 10k maps and holy cow, the 7800x 3d is over twice as fast as a 14900k. Over TWICE. That's insane. Problem is, they are both spitting fps into the hundreds, around 300-350 fps for the 14900k, and around 600-700 for the 7800x 3d.

Then you move into the actual big complicated maps of factorio and.....the x3d is even losing to alderlake and this time it actually matters cause it barely stays above 60. So yeah, what's the point of having a lead in light scenes when every cpu can spit lots of fps and then completely crumble in the harder scenes when you actually need the fps.

Especially since the game is basically restricted to 60ups, the goal on this one is to stay above 60 with as big a map as possible. The 3d chips fall flat on their face here. But then you watch hwunboxeds review and you have to pretend that they are 100% faster than the competition in this game, lol

https://factoriobox.1au.us/results?...9f36f221ca8ef1c45ba72be8620b&vl=&vh=&sort=ups

Same happens in eg. Cyberpunk and MSFS. Digital foundry made a video with live footage comparing the cpus in the actual hard scenes of those games and yeah, results were nowhere near close to what you'd expect. Reviews have the 7800x 3d hitting over 200 fps in cyberpunk, actual footage from a hard area (tom's dinner) and it drops to low 60s.... Literally 3 fps higher than the 9950x, a CPU that is unsuitable for gaming according to the experts here

Trying to derail again, so early.

The context given by the article is that the 9800x3d shows a marked improvement vs the 9700x. So much so that the benchmarks used indicate that the 9800x3d approaches the 14900/14700k. This is laudable given that the processor isn’t intended for the market that Puget serves.
Puget produce benchmarks that put their kit in the limelight with the best possible results FOR PUGET. As they mostly use Intel equipments surely it isn’t too much of a stretch to see that the benchmarks will favour Intel. (Board member notwithstanding)

The performance is achieved on fewer cores, that is a strong engineering achievement.

As for efficiency, wait for the reviews and proper power consumption results.

 

[TheHerald]

Trying to derail again, so early.

The context given by the article is that the 9800x3d shows a marked improvement vs the 9700x. So much so that the benchmarks used indicate that the 9800x3d approaches the 14900/14700k. This is laudable given that the processor isn’t intended for the market that Puget serves.
Puget produce benchmarks that put their kit in the limelight with the best possible results FOR PUGET. As they mostly use Intel equipments surely it isn’t too much of a stretch to see that the benchmarks will favour Intel. (Board member notwithstanding)

The performance is achieved on fewer cores, that is a strong engineering achievement.

As for efficiency, wait for the reviews and proper power consumption results.

Obviously those tests don't particularly scale with cores, so it isn't overly surprising. Actually as you can tell just by looking at the numbers the scores scale more with cache then with cores, so having the brand new 450$ R7 losing to last years 350$ i7 is indeed impressive. For the i7 part.

 

[stuff and nonesense]

Obviously those tests don't particularly scale with cores, so it isn't overly surprising. Actually as you can tell just by looking at the numbers the scores scale more with cache then with cores, .

the tests show intel scaling with cores and AMD 7000 doing similarly. The test with the 9800x3d shows good scaling against the 9700x, cache seems to work in some cases. What are you trying to prove?

 

[TheHerald]

the tests show intel scaling with cores and AMD 7000 doing similarly. The test with the 9800x3d shows good scaling against the 9700x, cache seems to work in some cases. What are you trying to prove?

How does it scale with cores? 14900k has 24c vs literally half on the 14600k (and much higher clockspeeds) and the performance difference is 15% barely. Clearly it's not scaling with cores. Cinebench scales with cores.

 

[Ogotai]

to buy any intel cpu, right now, is pointless.. why? can any one guarantee that intel fixed the stability and degredation issues of 13th and 14th gen,and those issues didnt carry over to the new chips ?

no one will know for sure for at least 6 months. no thanks. no one i know, is looking at intel now, 5 people i know were looking at upgrades, some were on the fence, others were amd or intel.. but now, thanks to those issues,all are going amd.. they just have no trust in intel now, amd probanly wont for 2 + years..

 

[TheHerald]

to buy any intel cpu, right now, is pointless.. why? can any one guarantee that intel fixed the stability and degredation issues of 13th and 14th gen,and those issues didnt carry over to the new chips ?

no one will know for sure for at least 6 months. no thanks. no one i know, is looking at intel now, 5 people i know were looking at upgrades, some were on the fence, others were amd or intel.. but now, thanks to those issues,all are going amd.. they just have no trust in intel now, amd probanly wont for 2 + years..

There is as much likelihood for a degradation issue as there is a chance the new x3ds will explode like the previous one. Close to none.

The fact that Intel increased warranty to 5 years means that the issue is fixed anyways. Unless you actually believe intel loves and cares about you and is willing to pay out of pocket to cover your cpu that will degrade. Let me assure you, they do not.

 

[Ogotai]

There is as much likelihood for a degradation issue as there is a chance the new x3ds will explode like the previous one.

do you have any proof of that, at all ? yea i dont think so, in any way.. i havent seen or read x3ds " exploding " even though they dont actually explode, for momths, have you ?

"The fact that Intel increased warranty to 5 years means that the issue is fixed

and point of that is what ? while you maybe ok dealing with rmas and the down time, others may not . and no it doesnt mean its fixed. nice try. no one knows for sure, for at least 6 months, period.

Unless you actually believe intel loves and cares about you and is willing to pay out of pocket to cover your cpu that will degrade. Let me assure you, they do not.

and that right there, is why they dont trust intel, and are going amd. so.. if intel wont pay out of pocket, seeing as you seem to be so confident in intel, how about, you pay for their board and cpu, and if they get any issues reguarding instability or degradation you repace it for them. no ? then they are going amd...

 

[TheHerald]

You sound very impartial so I'll take your word for it man. Thanks for the info.

 

[Ogotai]

You sound very impartial so I'll take your word for it man. Thanks for the info.

not impartial at all, just have zero trust in intel that that have fixed the issues with 13 or 14th gen, or if their new cpus dont have that issue...

but hey, if you are willing to pay for replacements, consider how great you claim intel is over amd, by all means.. im sure those i know would gladly take you up on that, cause they sure dont have the funds to take the risk buying intel right now, or for the next few years for that matter

 

[TheHerald]

Don't be harsh on yourself man, you really sound impartial. That's why I take what you are saying very seriously.

 

[Thunder64]

You sound very impartial so I'll take your word for it man. Thanks for the info.

Impartial? Coming from you? That was a joke that went over my head, right?