- Mar 3, 2017
- 1,747
- 6,598
- 136
The answer is no, and why would they show all of their cards already when they were patiently waiting for X Elite to go public for a proper comparison?
AFAIK Intel and AMD don't compare themselves to Apple in PR.
Intel, AMD and even Nvidia compare with Apple.AFAIK Intel and AMD don't compare themselves to Apple in PR.
I think Qualcomm are only doing so because they have literally been boasting that they had Axx µArch engineers working on Phoenix/Oryon.
MacOS still remains too separate from Windows or any serious level of compatibility to its software base on the level of Linux Proton/Wine, in no small part due to Apple's insistence on going their own way with Metal gfx API rather than the open standard of Vulkan, meaning that they cannot take advantage of the work done in DXVK and Zink.
Aged like fine wine in my opinion considering even AMD themselves say that Zen5 won't beat Zen4 X3D, and David's testing confirming that AMD is not sandbagging Zen5 at all lmao (in fact they're overhyping it, looking more like 6% for most use cases rather than 16%)Gotta admit, it'll be hilarious when the benchmarks come out and 14900K @ DDR5-8000/8400 is still ahead of the 9950X in gaming. Course, it'll be burning itself out in the process, but still.
Is that the game or not game stable version of the 14900k?Aged like fine wine in my opinion considering even AMD themselves say that Zen5 won't beat Zen4 X3D, and David's testing confirming that AMD is not sandbagging Zen5 at all lmao (in fact they're overhyping it, looking more like 6% for most use cases rather than 16%)
You're twice as likely to encounter a game that has no perceptible difference in performance on a 7800X3D than a 14900K/13900K with mainstream optimal setup configurations for both than encountering a game that has a noticeable difference.Is that the game or not game stable version of the 14900k?
I wish them luck, because they aint gonna get far chipping away at nVidia's lead.I bet their next target is the 3D rendering space something Nvidia has a huge lead in
That's a whole lot of not answering the question , stable or unstable defaults.You're twice as likely to encounter a game that has no perceptible difference in performance on a 7800X3D than a 14900K/13900K with mainstream optimal setup configurations for both than encountering a game that has a noticeable difference.
This is me summarizing HW Unboxed data.
"7800X3D is the best gaming CPU" is a overhyped meme. In reality, it is better than the alternatives for specific gaming scenarios.
did AMD design what is effectively Bulldozer 2: Electric Boogaloo?
Seems like your analysis is a bit flawed. If I see no perceptible difference between a 7800X3D and a 14900k (or what have you), why would I ever get the 14900k for a g4m3r rig? It uses a lot less power, is largely-insensitive to RAM speed (meaning I can save $$$ and time on that too), and basically is just a lot easier to build out for a competent g4m3r rig. In fact if 7800X3D stacks up too well against various Zen5 skus then I'm not really sure that g4m3rs are gonna want a Zen5 either. Until Zen5X3D comes out.You're twice as likely to encounter a game that has no perceptible difference in performance on a 7800X3D than a 14900K/13900K with mainstream optimal setup configurations for both than encountering a game that has a noticeable difference.
This is me summarizing HW Unboxed data.
"7800X3D is the best gaming CPU" is a overhyped meme. In reality, it is better than the alternatives for specific gaming scenarios.
Well AMD also measured a 12% lead for their Ryzen 5000XT refresh in a completely GPU Limited Scenario. Since ZEN4 (or at least RDNA3) you can't trust any of AMDs 1st party benches. Which is sad, because they were great at the beginning of Ryzen.In GB6 text processing he measure about 10% while AMD state that it s 19%.
In GB5 and AES XTS he measure about 12-13% while AMD state that it s 35%, so dunno what is the validity of his tests or if frequencies where accurate.
Raptor Cove in 14th Gen is at least +10% ahead in (geomean) gaming than vanilla Zen 4. Zen 5 vanilla will just barely* close that gap.Also the games that have a noticeable IPC advantage on GC compared to zen4 will likely be the games that see a good bump on zen5.
Raptor Cove in 14th Gen is at least +10% ahead in (geomean) gaming than vanilla Zen 4. Zen 5 vanilla will just barely* close that gap.
*Because-
1. In the overwhelming majority of cases, once you have enough cores, SMT hurts gaming performance.
2. Zen 5 behaves like a "8-wide" decode CPU only when you enable SMT. Otherwise, it is 4-wide.
We have no idea how 1 & 2 will affect gaming performance.
The chart says 8*2 bytes per cycle, the full L3 bandwidth is 32Bytes a cycle. So it is not impossible.You're not getting L1 bandwidth out of L3 region under any circumstances.
That's the point.
That's for data, max I've seen in nop chains is 17-18 bytes a cycle for a single core.The chart says 8*2 bytes per cycle, the full L3 bandwidth is 32Bytes a cycle. So it is not impossible.
Can happen with 2 parallel fetchesThat's for data, max I've seen in nop chains is 17-18 bytes a cycle for a single core.
And since Z5 makes no performance-relevant LLC changes...
...well it shouldn't happen.
That would pump L1/L2 fetch b/w even more sky-high, instead it sits even.Can happen with 2 parallel fetches
No, good per-fetcher will predict and fetch needed data from L2/L3/Memory. Since the bandwidth is higher than the core throughput, there nothing magical about it.That would pump L1/L2 fetch b/w even more sky-high, instead it sits even.
Makes no sense, haven't seen any core ever behave like that.
The pattern in question here gets prefetched anyway, you just run enough NOPs to roll L1/L2 capacity over.No, good per-fetcher will predict and fetch needed data from L2/L3/Memory. Since the bandwidth is higher than the core throughput, there nothing magical about it.
No again, ICache pre-fetcher will always fetch next stride from L2/L3. That is the case for a long time.The pattern in question here gets prefetched anyway, you just run enough NOPs to roll L1/L2 capacity over.
The weird thing is fetch b/w being even through all the cache level stride (where L3 should definitely be the Great Equalizer), either the prefetcher is way smarter and breaks his pattern or something weird is going on.
Considering an AMD rep also said it won't be faster than Zen 4X3D, it's unlikely to be much, if at all, faster than the 14900k.
Unless AMD are outright lying the 9950X should be around 10-15% faster than the 14900K on average.
You're twice as likely to encounter a game that has no perceptible difference in performance on a 7800X3D than a 14900K/13900K with mainstream optimal setup configurations for both than encountering a game that has a noticeable difference.
This is me summarizing HW Unboxed data.
"7800X3D is the best gaming CPU" is a overhyped meme. In reality, it is better than the alternatives for specific gaming scenarios.
They are not lying. They are using marketing like any other corporate entity.Unless AMD are outright lying the 9950X should be around 10-15% faster than the 14900K on average.
Since the 14900K now confirms what many of us suspected from the start - that running a CPU out of the box at settings which can only be called "Pentium 4 Extreme Edition deja vu" was asking for trouble - I cannot see why anyone would seriously have the unstable 14900K in any comparison since not even Intel know what the stable settings should be and a stock CPU which is not stable and regrades cannot be taken seriously.Unless AMD are outright lying the 9950X should be around 10-15% faster than the 14900K on average.