So RGT claims that the 9950X or whatever it is called is going to score ~170 pts in CineBench 2024 ST. Is it in line with expectations of folk around here?
That wasn't the my question - I asked about people's expectations with this "leaked" score of 170 pts, not directly related to claims from people with either a spotty track record at best or none at worst.The people around here are making claims regarding SPECint 2017 so not really comparable like that.
I presume those TPU perf charts are based off games. Those are not exactly good measuring tool of pure performance, too many other factors affecting those. Compute apps are. In something like Octane Render Titan X scores about 120 points. 4090 scores 1200, while having double VRAM (which is as important for rendering as the performance of the GPU chip itself - though its not reflected in that octane benchmark number).The 5930K launched late 2014, that would put it close to Titan X in early 2015. According to TPU perf charts, the 4090 is 5 times faster than the Titan X. Using the same ballpark system, 7950X is almost 3x faster than i5 10400. Factor in the clock advantage and some IPC improvements, and we're probably reaching 4x perf advantage over 5930K.
So, if 4090 is 5x faster for 80% more dollars, then the 7950x can be 4x faster for ~60% more dollars. The CPU should have launched for $960. /s
7950X is 5.55x faster than a 5930K in Cinebench R15.
| CPU | Geekbench 5 | Performance vs. 5930l |
|---|---|---|
| 5930k | 984 | |
| 5800x | 1728 | +76% |
| 7700x | 2202 | +124% |
| 7800x3d | 2137 | +117% |
| 7950x | 2215 | +125% |
The Zen 5 prediction sheetYour source please?
The point is that this logic does not apply to the Zen5 DT variants with the most cores only (like some have claimed), but for Zen5 in general.Budget-sensitive people won't be running for it but what's your point? If it's 40% faster then expect an absurd price. All that other stuff about more cores and types of work doesn't matter. If it's that far ahead of the competition they can charge whatever they want.
Flame bait removed.I don't think it will be even 25% faster and so will be cheaper than $800.
That's not how that works.Anyway, outside the US the economy is very weak. The USD is extremely strong. There's inflation, and salaries are not keeping up. Customers are very price sensitive. Zen5 will not sell well if AMD will bump the price. The customers are expecting stable or lower prices when comparing Zen5 models to corresponding Zen4 models.
The problem with your reasoning is that it applies to, say, any Z5 chip that's performance competitive. Which isn't a lot of them since again ST is king.Or they'll go with the competition from Intel, Qualcomm, et al.
Intel's net income is already at 1%. They can't compete any lower, or only with token discounts.Like others have mentioned, the competition will be fierce. Everyone want to keep or gain market share.
I think you are overestimating most people’s willingness to spend money on something as non-essential to life as faster CPU.That's not how that works.
If my choice is a $270 Z4 6 core that's 110% the performance of a $250 Z3 8 core, then I'm getting more bang for my money with the more expensive one.
If Z5 really is 40%+ more INT perf, it can afford to be massively more expensive and still sell. And ST is still king despite all the MOAR COARS crowd' screaming.
I use a 5600x. I'm not overestimating anything.I think you are overestimating most people’s willingness to spend money on something as non-essential to life as faster CPU.
Those same customers will also have the option of getting Zen 4 (or staying with one if they already have it). Why is it so hard to understand that AMD will price Zen 5 above Zen 4 until Zen 4 stocks run out? Then they will price Zen 5 at Zen 4 prices. Their first priority is to sell old stock. They can't just introduce a new CPU family and let it cause the old one to become worthless as crap. I'm not privy to their wafer supply agreement with TSMC but I bet they are STILL producing Zen 4 CPUs in some of the "older" TSMC fabs. Those CPUs need to sell out otherwise it's a total loss for AMD. So once more, in the beginning, Zen 5 stock will be unable to meet demand. That will keep prices high and some disappointed customers will buy Zen 4 or Intel or whatever coz that's all they can get their hands on or coz it's cheaper. Once the supply situation improves and Zen 4 stocks start dwindling, we will start to see Zen 5 sell at Zen 4 standard prices and Zen 4 will top the best deal of the week or month lists for a while until it is no more. Nothing too hard about that, is there? That's how the market works. Same thing happened with Zen 2 and Zen 3.The customers are expecting stable or lower prices when comparing Zen5 models to corresponding Zen4 models. Or they'll go with the competition from Intel, Qualcomm, et al.
OK, just promise not to be dishonest/hypocritical and not report buying a Zen 5 at launch and silently using it for 6 months before telling us hereEven 40 percent speed-up is not some be all and all, like its presented here, that everyone and their mother will run to buy money no object on day one, as if it somehow revolutionized computing. Its not quantum computer for gods sake. Its another incremental performance bump, that Most people wont be able to perceive in 95 percent of situations. And certainly wont feel the urge to randomly pay double for it than what they were used to, because apparently there is no competition.
Bit funny to read when you know that RDNA 4's biggest SKU will be 240mm² lol.The amd of Polaris is dead , it's time for some people to move on.
👍I'm all on the zen5 being a monster because amd have shown over 4 generations they can deliver high gains at the same sized /resourced core. So I believe they will further extact more performance per resource while making most resources pools significantly bigger.
And thus I expect pricing to align to that.
It kinda is. You're just trying to justified your own bias. In what universe is 40% not considered a large jump?Even 40 percent speed-up is not some be all and all, like its presented here
And how many amd products have been released besides Polaris and RDNA 3?
RDNA1 was good, and RDNA2 was even better. RDNA3 was bad - no doubt about it. The fact that AMD basically lied at the presentation made it worse.
Does it matter if people believe or not? People could not care less what may or may not have happened at AMD. Bottomline is that the product was not what it was hyped up to be.Some people still don't believe that RDNA 3 was an industrial accident it seems...
Again, even assuming the numbers are true
I may have phrased it wrong, the problem is not whether people believe in what RDNA 3 should've been. RDNA 3 had a flat promise of 50 + 50 in perf and efficiency.Does it matter if people believe or not? People could not care less what may or may not have happened at AMD. Bottomline is that the product was not what it was hyped up to be.
The facts of Z5 I do not know.The numbers may or may not be realistic but it is all fine and good knowing we are speculating. It might as well be greater, probably.
But when it gets dished out as facts it starts to get annoying.
This is why I say 40%+.You see, it is possible to have access to classified information but still acknowledge there are chances it may not be 100% true or simply acknowledge that not everyone accept everything as fact until everything is out.
I do believe quite a few. I find Kepler, Xino etc to credible and I wait and hope for things they say come to passSo unless you think this is all a giant troll/hoax and that Z5 is actually not a tock, not a large reshape, and is still on a 6 wide decode and we're all being Master Trolled by Kepler & Adroc, I'd say the odds are very, very, very high.