Rumor: AMD Ryzen 7000 CPUs would see maximum boost frequency of 5.85 GHz.

If the end top is around 5.8 GHz, it could be good for more budget CPUs.
In games single thread IPC is still king.

 

It's a little more complicated than that.
Just look at the Ryzen 7 5800X3D vs the Ryzen 7 5800X.

 

V-cache is the king

 

Cache is king, so is max tuned Ddr5 7000+
We want big cache and ultrafast tuned Ddr5.
From stock 4800c40 to tuned 7000c30, I gained 35% in min fps and average fps in BF 2042.

 

For a chiplet design that cache can matter a ton. The latency going out to main memory is higher for AMD vs Intel's monolithic designs.

 

sadly it will still be good but not "good enough" I think in the end.

 

The OS and scheduler can matter a lot too. For example, Linux benchmarks yield almost the exact opposite results as Windows, where games overall don't benefit that much from the cache, but productive workloads see a huge boost.

What else do you really want?

 

Education is important, but bigger numbers is importanter.

 

unquestionable dominance not a trading blows scenario. if it's not unquestionably better the next iteration Intel is just waiting to release will be "better" thus winning the 'mah FPS r bettr" argument.

 

It is a choice between high cache or high mem speed.
Cache can boost 30% and memory can boost 30%, but high cache and memory speed at the same time will maybe only boost a combined 35% and not 30 + 30 = 60.

AMD was stuck on 3800 DDR4 so they invented a workaround for the "slow" memory.
Intel is not able to mount cache to the 12900k because the worse heat transfer would cripple the chip to much, so high mem speed is needed.

 

Well there's 2 ways to interpret that:
1. If such a massive leap was as feasible as you think, then what's preventing Intel from doing the same?
2. AMD is a business. If they throw everything they've got at making the best chip possible, there won't be much room left for improvement in next generations. At that point, people would continue to expect more of such great improvements, which obviously won't happen. That means AMD sees fewer sales after the "big improvement", and might actually struggle to continue R&D.
Think of it like a biological virus: it's "goal" is to reproduce. It has to be contagious enough to reliably spread to other organisms, but if it's too contagious then all possible hosts become infected and the virus goes extinct. Developing a product works the same way: each iteration needs to be good enough to compel consumers to upgrade, but not so good that it becomes the only upgrade. For most of us, the improvements are incremental enough that we skip a generation or two, but with everyone staggering their upgrades, that makes for a relatively steady income.

 

that's just it. I don't think they are able to do it. Hence why I'm saying it will be "better" but not by enough to be revolutionary

 

I hope they learned from the past fiasco where the bald guy gave an unrealistic example and people took it as literal attainable stock frequency ending up with AMD removing the number and just leaving the word GHz. Was rather funny though

 

Systems not so future proof when AMD is around.

 

More cache on 13900k, so it wil be fun to see Can't wait to OC 13900k and 7 series AMD with fast DDR5

 

At launch, Lisa Su said the TDP was 170W, IIRC. Didn't say anything about socket power, but as that is a fixed ratio of 1.35X the TDP, it's not a surprise as the socket power for all AMD CPUs is figured that way, according to Hallock, and for an excellent reason: AMD motherboards will presumably be able to house some AMD CPUs past the 7000 series, etc. You always want to leave a little headroom. Robert Hallock interjected mistakenly that the 170W was the socket power. This is just not a big deal at all--Hallock made a simple error and corrected it the next day.

Hallock also stated that the 170W TDP would be for the highest end of the Desktop 7000 CPUs, that AMD would still be employing 65W and 105W CPUs with AM5. Hot Hardware had a nice interview with Hallock where he went over all of that stuff the day after launch. Additionally, Hallock said the engineering sample was not tapped out at 170W TDP during the demo, so 5.85GHz sustained is certainly believable.
Nobody knows things for certain, though, until they ship this fall, of course.

 

It is 0,75MB extra L2 cache per core, it is high amount extra, but the P core L3 is still only 20MB + 16MB E core cache (according to a leaked CPU-Z screenshot)
A 5600x still has more L3 cache then the P cores part of the 13900k, I guess that is why Intel likes fast memory so much.

 

I saw those tests from phoronix. Dues Ex had a 36% speed up in 1080p on LInux. I think its the game selection they used that didn't gain much from the larger cache. The productivity side did better in Linux but then doesn't it normally.

 

I really want to get my hands on one. But I have 2 gaming PC's and a laptop already.