• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

AMD Delays Launch of Ryzen 9000 Series Processors

they're waiting for intel patches to not YOLO 2 cores at 6.2Ghz and 1.6v so that they can get that single core win.

You're not finding quality issues 7 days before launch, on the same day that your competitor announces a patch in 'mid august' - that you can magically address in 15 additional days.

At this point all the pallets are shrinkwrapped and ready to go. They had to sail those dies from Taiwan, nothing is happening in 15 days - you couldn't fix a typo on the box in that time.
I believe this is a big part of it.

Apparently everything is being recalled, including OEM CPUs sent to system integrators that may already be sat in motherboards in prebuilt systems already. Presumably 15 days is how long it'll take for AMD to send a new batch of more-carefully-tested CPUs by boat out to "the channel" and any extra delays are going to impact the retailers and system integrators - so it'll be more of a paper-launch than usual as the launched products might still be in transit, or at least won't be ready for sale in a prebuilt until the system integrator's had a chance to install the replacement CPUs and re-run their pre-ship tests again.

It's awfully convenient for AMD that this is after Microsoft's microcode that's backing off the ridiculous factory overclocks on the 14th Gen. It's highly likely that AMD were on the fence about delaying and this mis-step by Intel was an opportunity AMD couldn't resist as a win-win scenario that may well hand them a bigger benchmark victory and make AMD look like the more careful, trustworthy, reputable CPU vendor.
 
There is even palettes in retailers ready for orders. I dont like to speculate, but the time frame just seems so short and really coincidental with the schedule for Intel's new microcode.
The reason they gave was very vague, and the timeline to fix was ridiculously short for any serious redistribution, extremely specific, and identical to the timeline given by intel... I mean - i'm not a biologist but....
 
I think the hyper threading was revolutionary in terms of added performance, when it hit the market. it did boost performance exponentially. Could call that revolutionary. Now it is bullshit but ecore vs pcore scheduling is revolutionary?
Mixing 2 cores (not architectures) is revolutionary? Is it though? These ecores are literally cut down older cores Intel used previously in a product. I'm sorry, but that is not revolutionary to me. What would have been revolutionary if you put x86 arch cores and Arm arch cores together. That could've been revolutionary.
Fair point. I guess it was kind of revolutionary to me, as I've seen nothing like it before (not on desktop at least). Although, as I said, due to software schedulers and Windows 11, the concept never convinced me that it's something I'd ever want.
 
traditional oldschool homogenous gaming performance with no thread scheduling bullshit.
This^

We're on desktop. All them E-cores (ZENc ones as well) belong to the mobile segment.
 
This^

We're on desktop. All them E-cores (ZENc ones as well) belong to the mobile segment.
IDK - if they can get it right, the e-cores and especially zenC arfe stronger than HT cores by a good bit, and they allow more aggro prefetching and fewer security nanny issues -- I could see it working. They just came out with an e-core only xeon that absolutely slaps in MT.
 
Ecores and ccds are different solutions to the same problem. How to increase performance per die without increasing man costs. They both have strengths and weaknesses but for my personal tastes ecores are better than splitting the cores on 2 ccds.

I don't like / want 16 ecores (8 are fine) but I don't like cores split on 2 ccds either. So I'm probably skipping both for now until amd decides to move to 12core single ccds with 3d on top, or intel increases their pcore count.

But let's not pretend amds solution is better, it's not. They both have their issues for now. I'll admit though amd seems closer to the perfect solution (the 12core chip I mentioned above).
You focus on irrelevant things instead of looking at the metrics that give you a perspective of the product. For me, it does not matter if the CPU has CCD's or not. You dont want 16 cores, accusing AMD for stagnation (which is ridiculous in my opinion) and advocate for increasing ecores. I dont want this and, that getting into dislike the CCD config, that doesnt mean crap if the performance is there. I really dont think, ecores and CCD are solving the same problem. Maybe along the lines you can say that but that is not the main purpose of those. In my eyes ecores are for lower power consumption without loosing too much performance. Nobody is pretentding anything about AMD's solution. They have the solution and increased cores to 16 in the desktop segment. You have server segment were cores matter and the number of cores there is staggering with AMD products. ecores are ecores only not real performance cores. Never will be.
 
You focus on irrelevant things instead of looking at the metrics that give you a perspective of the product. For me, it does not matter if the CPU has CCD's or not. You dont want 16 cores, accusing AMD for stagnation (which is ridiculous in my opinion) and advocate for increasing ecores. I dont want this and, that getting into dislike the CCD config, that doesnt mean crap if the performance is there. I really dont think, ecores and CCD are solving the same problem. Maybe along the lines you can say that but that is not the main purpose of those. In my eyes ecores are for lower power consumption without loosing too much performance. Nobody is pretentding anything about AMD's solution. They have the solution and increased cores to 16 in the desktop segment. You have server segment were cores matter and the number of cores there is staggering with AMD products. ecores are ecores only not real performance cores. Never will be.
E-cores are for increasing core count without significantly increasing die area (I don't think power consumption has got a lot to do with it). CCDs are for increasing core count without significantly increasing manufacturing costs. Neither of them are meant for people whose application performance is hurt by inter-core latency or software scheduling. They are meant for people looking for multi-threaded performance and nothing else.
 
In my eyes ecores are for lower power consumption without loosing too much performance. Nobody is pretentding anything about AMD's solution.
But what it is in your eyes is irrelevant, isn't it? 8Pcores at 65w (just an example) are faster than 8ecores at 65w, therefore OBVIOUSLY ecores have nothing to do with power.

E-cores are for increasing core count without significantly increasing die area (I don't think power consumption has got a lot to do with it).
Exactly.

You have server segment were cores matter and the number of cores there is staggering with AMD products. ecores are ecores only not real performance cores. Never will be.
Well great, but I don't care about server products. At the 300 to 400$ segment they are stuck at 6 and 8cores for how many years now? 7?
 
Well great, but I don't care about server products. At the 300 to 400$ segment they are stuck at 6 and 8cores for how many years now? 7?
That's because that segment is directed at home users and gamers who still don't need more than 8 cores. On the other hand, we've been getting some decent performance increases across those 8 cores, which is exactly what a gamer needs, imo.
 
That's because that segment is directed at home users and gamers who still don't need more than 8 cores. On the other hand, we've been getting some decent performance increases across those 8 cores, which is exactly what a gamer needs, imo.
That's not a reason. If you don't need 8 cores and you are fine with 6 youll buy the 6 core part of course, but it doesn't need to be priced at 299. Did people need more cores back in 2017 or 2018? Cause let me remind you, you could grab the R5 1600 for 80$ in 2018-2019. Now we are in 2024 and the 7600x, 2 years later is over 200$.


I'm not gonna argue whether 6 and 8 cores are fine - that's down to your usecase I guess - but the point is you can't have your baseline model starting at 6 cores and 299$. Now some rumors about zen 5 are hinting at a 229$ price for the 6core chip which is okay. Not great, but it's a step forward.
 
That's not a reason. If you don't need 8 cores and you are fine with 6 youll buy the 6 core part of course, but it doesn't need to be priced at 299. Did people need more cores back in 2017 or 2018? Cause let me remind you, you could grab the R5 1600 for 80$ in 2018-2019. Now we are in 2024 and the 7600x, 2 years later is over 200$.


I'm not gonna argue whether 6 and 8 cores are fine - that's down to your usecase I guess - but the point is you can't have your baseline model starting at 6 cores and 299$. Now some rumors about zen 5 are hinting at a 229$ price for the 6core chip which is okay. Not great, but it's a step forward.
I don't disagree with the sentiment of cheaper hardware. I'd like to get mine at 2016 prices, too, even if adjusted for inflation. But unfortunately, that's not the world we live in today when there's price gouging on all sides, including at TSMC and other chip manufacturers.
 
I don't disagree with the sentiment of cheaper hardware. I'd like to get mine at 2016 prices, too, even if adjusted for inflation. But unfortunately, that's not the world we live in today when there's price gouging on all sides, including at TSMC and other chip manufacturers.
That would be correct but the price increase happened way before the covid inflation bs. Zen 3 basically didn't introduce the cheaper non X variant on launch which resulted in literally a 50% increase between zen 2 and zen 3 on the same core count chips. If i'm not mistaken 3600 launched at 199$, 5600x a year later launched at 299$.

But anyways, pray to god the rumors are correct pricing on zen 5 looks a big more tolerable. 229 and 299 for the 6 and 8 core part ain't atrocious.
 
But what it is in your eyes is irrelevant, isn't it? 8Pcores at 65w (just an example) are faster than 8ecores at 65w, therefore OBVIOUSLY ecores have nothing to do with power.
Because you should be judging a product for what it offers so I it is " not in my eyes" or a personal preference thing but CPUs performance not how many CCD's it has. That is ridiculous in my opinion. It's like judging a car for its color. You may have a preference for features but not something, that impacts nothing and the product is one of the best in the market.
That is just my opinion but you do you. Shape of a processor may matter to you too but does it really change things?
 
That would be correct but the price increase happened way before the covid inflation bs. Zen 3 basically didn't introduce the cheaper non X variant on launch which resulted in literally a 50% increase between zen 2 and zen 3 on the same core count chips. If i'm not mistaken 3600 launched at 199$, 5600x a year later launched at 299$.
At that point they had objectively the best products, and could increase the prices to match that fact. It’s a company, like intel, trying to make money, not some charity.
 
Because you should be judging a product for what it offers so I it is " not in my eyes" or a personal preference thing but CPUs performance not how many CCD's it has. That is ridiculous in my opinion. It's like judging a car for its color. You may have a preference for features but not something, that impacts nothing and the product is one of the best in the market.
That is just my opinion but you do you. Shape of a processor may matter to you too but does it really change things?
Are you saying multiple CCDs have the same drawbacks as the shape of the processor? Okay man.

At that point they had objectively the best products, and could increase the prices to match that fact. It’s a company, like intel, trying to make money, not some charity.
And I'm a consumer so why would I defend said company when it literally goes against my own benefit?
 
That would be correct but the price increase happened way before the covid inflation bs. Zen 3 basically didn't introduce the cheaper non X variant on launch which resulted in literally a 50% increase between zen 2 and zen 3 on the same core count chips. If i'm not mistaken 3600 launched at 199$, 5600x a year later launched at 299$.
Do you remember what was the price for an 8 core Intel counterpart was or 6c? Need I remind you? You compare x to non x. non x are always cheaper. Not convinced sorry.
Are you saying multiple CCDs have the same drawbacks as the shape of the processor? Okay man.
:D sure bro. Now try reading it backwards maybe you will get it.
 
Last edited:
Nvm, let's stick to zen 5
 
Last edited:
Just typo:笑:
 

Attachments

  • 726.png
    726.png
    115.5 KB · Views: 57
Just typo:笑:
Ian probably just meming. There would be no point in having SI’s strip down already built systems, considering that the packaging is not even provided to them. Some sources indicated weird performance issues on select few processors in gaming workloads, which sounds a lot more reasonable.
 
IDK - if they can get it right, the e-cores and especially zenC arfe stronger than HT cores by a good bit, and they allow more aggro prefetching and fewer security nanny issues -- I could see it working.
That's a big if though.

Will observe ARL and BTL launches to gather more knowledge on the matter.
 
E-cores are for increasing core count without significantly increasing die area (I don't think power consumption has got a lot to do with it). CCDs are for increasing core count without significantly increasing manufacturing costs. Neither of them are meant for people whose application performance is hurt by inter-core latency or software scheduling. They are meant for people looking for multi-threaded performance and nothing else.
Bingo. You are the first person mentioning the legacy issue here.
 
Ian probably just meming. There would be no point in having SI’s strip down already built systems, considering that the packaging is not even provided to them. Some sources indicated weird performance issues on select few processors in gaming workloads, which sounds a lot more reasonable.
GTlTXJ5bAAEITbD.png

Ryzen 9 9700X instead of 7
 
A wise desision from AMD's side, if there are any doubt on cpu's stability and function. A delay is apselutely better than a huge mess of defective cpu's after launch. That will only hurt amd image negatively.

We have just seen the mess intel has been in and still is in with there gen 13/14 cpu's. But we must not forget either the early state of amd zen 4 launch where some cpu's litterly burned up or melted perhaps even self destructed.

Zen 4 and Intels latest gens problems. Kind of confirms my worries/suspicion on that in the hunt to be the one to have the fastest cpu over the competiter. Amd and intel pushing the cpu's to far all ready at stock and lower the lifespan or degrade them to fast. We have seen Intels and amd latest chip runs hot and consume a high amount of power. Off cause naughty motherboard venders running the cpu out of spec as stock dosent help the situation either. But a cpu running at 95-100 C at high load can't be healthy. Despite intel/amd claims it by design.
Laptop CPUs have been regularly running 90+ degrees reliably for over 15 years now. I think the engineers know better then us at what temps these CPUs can run at.
 
Back
Top