AMD Ryzen 7 9800X3D

[Vayra86]

Well we don't really know how good it is in games thay can actually benefit.

There are some games in the bench suite that show a real gap... but it still gets there with a lot more power. Even if you get the best case scenario games only though, its what, 12%? As far as I'm concerned the only real difference here is the higher clocks. It sits a good 200-400mhz above a 7800X3D in games, and that's also about what we're seeing in perf, plus or minus a few %.

Other than that there is no reason to believe there is more magical sauce here. You get performance, if its not necked otherwise, through higher clock.

 

[LittleBro]

Source: HUB

How come HUB's results are so much different to W1zzard's? Did they use Win10?

 

[oxrufiioxo]

Dude..

I just looked at the review.

It trounces my 5900X, and 5800X3D in absolutely everything.

Good deal.

Yeah, it's slightly better than I expected it to be with only the power consumption being a minor bummer but fine given the performance level.

I remember when Intel was 5% ahead at gaming and people acted like it was a bloodbath lol now it's not even remotely close....

 

[dir_d]

Im holding on to my 5800X3d but if anyone on this platform was itching to upgrade i think the 9800X3d is it. I will probably go AM5 when its got some crazy Microcenter bundle deal.

 

[Vayra86]

Power consumption sucks... disappointing. I didnt think it would be that much higher, % wise. We all know anything under 150 is nothing. Still upsetting. Next gen stuff will prolly need more too. Its gonna end up like Intel...

It has a long way to go until then, but yeah, only smaller nodes will save the current design from getting worse efficiency at this point.

 

[Baba]

I've got a 7700X and game at WQHD, I believe the 9800X3D would give me about a less than 10% boost on average. I think I'll sit this one out. Hopefully, AMD has one or two more AM5 CPUs incoming, otherwise the whole AM5 longevity thing will be pointless.

My 5600 is only 7.2% behind this king of gaming performance at 4k. I don't know if that makes me happy or sad. It's really hard to build a whole new system just for that gain. I have no hope that the next AM5 CPU will be much better. This platform will end up a dissapointment. I don't care about longetivity if they keep relesing CPUs with minimal gains.

 

[oxrufiioxo]

How come HUB's results are so much different to W1zzard's? Did they use Win10?

24h2 and I believe they also used pbo/CO so people should probably drop 2-3%

Also testing is done in different sections of the games and games are highly dynamic.

 

[Vayra86]

Yeah, it's slightly better than I expected it to be with only the power consumption being a minor bummer but fine given the performance level.

I remember when Intel was 5% ahead at gaming and people acted like it was a bloodbath lol now it's not even remotely close....

Things look better when you're expecting to get disappointed hehehe

 

[phints]

Ok as expected 9800X3D is a gaming performance beast (and quite good at productivity). It's a shame it lost some of that beautiful power efficiency of the 7800X3D, but it's still well above Intel there too.

Only cons here are the very high price, and 20W higher idle power consumption than Intel.

 

[Frick]

Benchmarking and its methods are a total mystery even to moderators, apparently.......

You benchmark at lower resolutions to push the CPU to the limit to test its actual capability. Same reason you test a GPU at 4k even if it cannot push 4k properly. A CPU test at 4k is as pointless as a GPU test at 720p. This gets explained every time there is a CPU benchmark and people still cant comprehend it.

Also, high refresh rate gamers? You know, the guys who do esports and competitive shooters? They all play at 1080p, and they want HIGH framerates. So its not a useless test anyway.

But few test those games (apart from CS2 maybe). It's AAA games or at best MS Flight Sim and they run those at 1080P, which tells you nothing of how fast it realistically is in games where the cache stuff and new architecture actually matters.

 

[Baba]

Im holding on to my 5800X3d but if anyone on this platform was itching to upgrade i think the 9800X3d is it. I will probably go AM5 when its got some crazy Microcenter bundle deal.

Same here. I have a 5600. The only thing of worry is whether PCIe Gen3 will hold back a 5080/5090 enough to justify an upgrade.

 

[oxrufiioxo]

Things look better when you're expecting to get disappointed hehehe

Zen5 vanilla well that kinda sucks at gaming... Arrow Lake hold my beer..... Definitely tempered my expectations lol

 

[AVATARAT]

Ok, I watched the yt reviews and enjoyed the bloodbath.
Now the serious question. Should the 5800X3D owners upgrade or not?
The Hogwarts Legacy numbers have been haunting me since 7800X3D came out but now it's getting out of hand.

4080/7900 XTX++ cards, yes if they want some % more.

 

[Frick]

There are some games in the bench suite that show a real gap... but it still gets there with a lot more power. Even if you get the best case scenario games only though, its what, 12%? As far as I'm concerned the only real difference here is the higher clocks. It sits a good 200-400mhz above a 7800X3D in games, and that's also about what we're seeing in perf, plus or minus a few %.

Other than that there is no reason to believe there is more magical sauce here. You get performance, if its not necked otherwise, through higher clock.

I'm talking sim games here, like late game Farthest Frontier or MS Flight Sim. Or as someone mentioned, long term Factorio. That's what I want numbers on.

 

[freeagent]

I know is is for AMD.. but I hope Intel can bounce back.

The back and forth is exciting..

 

[Vayra86]

I'm talking sim games here, like late game Farthest Frontier or MS Flight Sim. Or as someone mentioned, long term Factorio. That's what I want numbers on.

Yes. That would be sweet. @W1zzard take note

 

[btk2k2]

24h2 and I believe they also used pbo/CO so people should probably drop 2-3%

Also testing is done in different sections of the games and games are highly dynamic.

They used 24H2 (except for the Arrow Lake parts). They did pbo/CO testing as an extra bit of info but that is not part of their main charts so no, you should not drop 2/3%,

 

[Carillon]

Well, now there's no reason for the dual CCDs SKUs to come with just one 3d v-cache die.

 

[praescepter]

How come HUB's results are so much different to W1zzard's? Did they use Win10?

Better custom scenes. Look at the Hogwarts numbers for example. The HUB review is better in the gaming part.

 

[AnotherReader]

Thanks for the review. It's interesting to see that contrary to the received wisdom, extra cache helps workloads such as encoding, rendering, and game development as well. Even scientific workloads see an improvement.

 

[oxrufiioxo]

They used 24H2 (except for the Arrow Lake parts). They did pbo/CO testing as an extra bit of info but that is not part of their main charts so no, you should not drop 2/3%,

I just skimmed the results and seen the C/O slide thanks for clarifying that.

Somthing I didn't mention above but I believe HUB also uses the most cpu limited section they can find. I could be wrong but I think TPU chooses more of a middle ground section. Which can also have a large impact on numbers.

 

[mb194dc]

So minimal gains at normal use case resolutions 1440p+ v 3 generations back. Will be less on tuned 58 or 57 3D presumably as well.

This with a 4090 which has <1% market share.

It's better than Intel, still marginal gains for the median use case for a lot of money.

 

[KarymidoN]

Techpowerup is kidding, right?

They are testing AMD with Win23H2....
so no BRANCH PREDICTION PATCH

it seems Hardware Unboxed is the only one testing AMD with Windows 24H2...

So minimal gains at normal use case resolutions 1440p+ v 3 generations back. Will be less on tuned 58 or 57 3D presumably as well.

This with a 4090 which has <1% market share.

It's better than Intel, still marginal gains for the median use case for a lot of money.

calm down ppl, @W1zzard will do his usual followups, just chill and let the man WORK! 24H2 tests will come, just wait, AMD said 24H2 rn is kinda buggy, just wait.

 

[HD64G]

That CPU will become the new king in PC market. Will help also to lower the price of the previous gens. Good job there AMD engineers in general and marketing department (for its price being sensible).

 

[Dredi]

So minimal gains at normal use case resolutions 1440p+ v 3 generations back. Will be less on tuned 58 or 57 3D presumably as well.

This with a 4090 which has <1% market share.

It's better than Intel, still marginal gains for the median use case for a lot of money.

No can do until nvidia ups their game. In many games with fps oriented settings this does produce better results in a meaningful way. Max settings 4k not so much, but when has that _ever_ happened?