NKD wrote on Oct 9, 2024, 07:21:
CPUs have always been reviewed at lower resolutions because people want to know how much headroom they have to not be bottlenecked by it in the future. Just because two CPUs give you the same FPS at 1440p or 4k now, doesn't mean that'll be true when you buy a new GPU next year.

Interesting. Out of curiosity, I tried this technique:

Comparing AMD 3700X to Intel 11900K:

• 2020 review from TPU (using 3080 10GB) showed negative performance deficit from 11900K to 3700X as:
  -20% 720p
-14% 1080p
-9% 1440p
-2% 4K
• 2024 review from TPU (using 4090 24GB) showed negative performance deficit from 11900K to 3700X as:
  -19% 720p
-18% 1080p
-12% 1440p
-5% 4K

I think that aligns with your hypothesis, even with different GPUs and games being tested (although Cyberpunk is in both). The performance gap from 11900K to 3700X has widened by 3-4% over those 4 years (except at 720p) as the 11900K is better at leveraging a much faster GPU with newer titles.

But I think the margins are still too small to matter, compared to the gains offered with a GPU upgrade. The 3700X user would have been better off upgrading to 3080 or 3080ti in 2021 instead of a new CPU, for gaming at 1080p or higher. And I think that rings true today with Ryzen 7000/9000 series and Intel 13th/14th gen... most gamers are better off getting discounted and bundled Ryzen 5000 or Intel 12th gen and the fastest graphics card they can buy.

This comment was edited on Oct 9, 2024, 13:43.

Why are y'all engaging with a tryhard edgelord? Seriously, the individual never engages in good faith or with intellectual honesty.

VaranDragon wrote on Oct 9, 2024, 04:03:

RogueSix wrote on Oct 8, 2024, 17:19:
* REMOVED *

You are the clown, those youtube sites he mentioned happen to be some of the most honest and informative you can find (especially Gamers Nexus) But hey, keep on spreading the lies around, it's all the rage these days.

No one blasphemes Tech Jesus!

This comment was edited on Oct 9, 2024, 15:57.

theglaze wrote on Oct 8, 2024, 19:51:
I just think it's dumb to use 720p or 1080p benchmarks for high fps results, just to show a difference between the processors. That resolution only seems relevant to gaming titles used in the competitive scene.

If the story ends up being "if you game at 1440p or higher, there's no advantage to these new and expensive CPUs over last gen", then maybe that's the advice that reviewers should give.

Go buy a much cheaper 'old' CPU bundled with RAM+mobo at a discount, and put aside the savings to pay Nvidia

CPUs have always been reviewed at lower resolutions because people want to know how much headroom they have to not be bottlenecked by it in the future. Just because two CPUs give you the same FPS at 1440p or 4k now, doesn't mean that'll be true when you buy a new GPU next year.

RogueSix wrote on Oct 8, 2024, 17:19:
* REMOVED *

You are the clown, those youtube sites he mentioned happen to be some of the most honest and informative you can find (especially Gamers Nexus) But hey, keep on spreading the lies around, it's all the rage these days.

This comment was edited on Oct 9, 2024, 15:56.

I just think it's dumb to use 720p or 1080p benchmarks for high fps results, just to show a difference between the processors. That resolution only seems relevant to gaming titles used in the competitive scene.

If the story ends up being "if you game at 1440p or higher, there's no advantage to these new and expensive CPUs over last gen", then maybe that's the advice that reviewers should give.

Go buy a much cheaper 'old' CPU bundled with RAM+mobo at a discount, and put aside the savings to pay Nvidia

This comment was edited on Oct 8, 2024, 20:03.

RogueSix wrote on Oct 8, 2024, 17:19:
* REMOVED *

You can troll better than that, I've seen it. Gotta put in the effort.

This comment was edited on Oct 9, 2024, 15:55.

* REMOVED *

This comment was deleted on Oct 9, 2024, 08:13. Reason: Misquoting (rule 3)

jdreyer wrote on Oct 8, 2024, 15:57:
No more Anandtech.

No, but Anandtech hasn't really been good for years, not since it was sold and Anand left. I tend to stick with Gamers Nexus, Hardware Unboxed, and JayzTwoCents for most stuff these days.

Nullity wrote on Oct 8, 2024, 14:26:

RogueSix wrote on Oct 8, 2024, 14:05:

Nullity wrote on Oct 8, 2024, 13:45:
... and for gaming they compared to the 7950X3D instead of the 7800X3D, showing they're still not attempting to be honest

Huh? Depending on the game/website/benchmark/methodology, the 7950X3D is actually a tiny little bit faster than the 7800X3D but the performance difference is generally negligible. It is also the 16 core part and the Intel CPU is a 24 core part so actually I would say that it is more honest to compare 16 cores (7950X3D) vs. 24 cores (285K) than 8 cores (7800X3D) vs. 24 cores (285K).

The 7950 has 2 CCDs (each with 8 cores), only one of which uses the X3D cache. If the non-3D cache CCD isn't disabled during gaming (core parking), performance will be worse than the 7800 with just a single CCD. So in essence (for gaming specifically), the 7950 and 7800 X3D CPUs are both 8-core parts. I don't trust Intel to do this comparison properly.

Perhaps I should have gone into more detail in my original post. In any case, obviously we need to wait for independent tests.

No more Anandtech.

Nullity wrote on Oct 8, 2024, 14:26:

RogueSix wrote on Oct 8, 2024, 14:05:

Nullity wrote on Oct 8, 2024, 13:45:
... and for gaming they compared to the 7950X3D instead of the 7800X3D, showing they're still not attempting to be honest

Huh? Depending on the game/website/benchmark/methodology, the 7950X3D is actually a tiny little bit faster than the 7800X3D but the performance difference is generally negligible. It is also the 16 core part and the Intel CPU is a 24 core part so actually I would say that it is more honest to compare 16 cores (7950X3D) vs. 24 cores (285K) than 8 cores (7800X3D) vs. 24 cores (285K).

The 7950 has 2 CCDs (each with 8 cores), only one of which uses the X3D cache. If the non-3D cache CCD isn't disabled during gaming (core parking), performance will be worse than the 7800 with just a single CCD. So in essence (for gaming specifically), the 7950 and 7800 X3D CPUs are both 8-core parts. I don't trust Intel to do this comparison properly.

Perhaps I should have gone into more detail in my original post. In any case, obviously we need to wait for independent tests.

Yes, but as you can see from the benchmarks (on the same slide), they also compared the two CPUs outside of gaming so I think that Intel chose the correct competitor. The 7950X3D and 7800X3D are almost identical in games due to the reasons you stated but in apps all 16C/32T of the 7950X3D get to work so a comparison against this specific AMD CPU seems fair. In fact, wrt gaming, good on Intel to compare to the 3D parts. They could have compared to the regular non-3D CPUs to make themselves look better in gaming if that was the idea (my Chinese is kinda rusty so I can't read the slide titles ).

Oh well, only 16 more days... then we'll get independent benchmarks of 285K vs. 7800X3D, 7950X3D, 14900K(S), 9700X, 9900X, 9950X until we're blue, green and pink in the face .

RogueSix wrote on Oct 8, 2024, 14:05:

Nullity wrote on Oct 8, 2024, 13:45:
... and for gaming they compared to the 7950X3D instead of the 7800X3D, showing they're still not attempting to be honest

Huh? Depending on the game/website/benchmark/methodology, the 7950X3D is actually a tiny little bit faster than the 7800X3D but the performance difference is generally negligible. It is also the 16 core part and the Intel CPU is a 24 core part so actually I would say that it is more honest to compare 16 cores (7950X3D) vs. 24 cores (285K) than 8 cores (7800X3D) vs. 24 cores (285K).

The 7950 has 2 CCDs (each with 8 cores), only one of which uses the X3D cache. If the non-3D cache CCD isn't disabled during gaming (core parking), performance will be worse than the 7800 with just a single CCD. So in essence (for gaming specifically), the 7950 and 7800 X3D CPUs are both 8-core parts. I don't trust Intel to do this comparison properly.

Perhaps I should have gone into more detail in my original post. In any case, obviously we need to wait for independent tests.

Nullity wrote on Oct 8, 2024, 13:45:
... and for gaming they compared to the 7950X3D instead of the 7800X3D, showing they're still not attempting to be honest

Huh? Depending on the game/website/benchmark/methodology, the 7950X3D is actually a tiny little bit faster than the 7800X3D but the performance difference is generally negligible. It is also the 16 core part and the Intel CPU is a 24 core part so actually I would say that it is more honest to compare 16 cores (7950X3D) vs. 24 cores (285K) than 8 cores (7800X3D) vs. 24 cores (285K).

theglaze wrote on Oct 8, 2024, 13:55:

Nullity wrote on Oct 8, 2024, 13:45:
That said, there doesn't really seem to be any improvement over the 14th gen

Power consumption seems to be the biggest improvement, maybe -33% less power required for similar performance results as last gen?

Which means the Intel tuners are going to push these new chips to show a bigger advantage with OC and toss the power limits out the window. Exposing themselves to degradation, assuming Intel has not already baked that in at default speeds

You're right, I should have mentioned that, I was focused on performance. The new CPUs do appear to improve power efficiency, which is certainly the biggest thing Intel needed to work on.

Nullity wrote on Oct 8, 2024, 13:45:
That said, there doesn't really seem to be any improvement over the 14th gen

Power consumption seems to be the biggest improvement, maybe -33% less power required for similar performance results as last gen?

Which means the Intel tuners are going to push these new chips to show a bigger advantage with OC and toss the power limits out the window. Exposing themselves to degradation, assuming Intel has not already baked that in at default speeds

First Gaming Benchmarks for Intel Core Ultra 9 285K Leaked - DSOGaming.

"Leaked"... by Intel. You're not fooling anyone. You've conditioned the collective tech industry to laugh out loud at you when you release anything even resembling benchmarks or performance measurements.

That said, there doesn't really seem to be any improvement over the 14th gen (anyone who criticized AMD for the 9000X CPUs, keep that in mind), and for gaming they compared to the 7950X3D instead of the 7800X3D, showing they're still not attempting to be honest. (Not that I'm claiming AMD are saints, but they have a better track record.)