AMD Ryzen 9 4950X "Vermeer" Tested, the Sample Boosts to 4.8 GHz

It's not chip designs that are the issue -- it's the fabs to produce them.

That's the thing, though, when they were developing Pentium M alongside Pentium they didn't have to wait around and test different fabs. Pentium M was established for years at that point and more than ready to take the lead when Pentium choked.
 
It's nice to see them get near that 5.0 Ghz mark and would be even nicer to see it hit but if IPC improves it's mostly symbolic and completely meaningless, we all know how wonderful Piledriver was at 5.0 Ghz. I'm hoping to see them close the remaining gap in gaming performance and I think it's fairly likely.

I don't think they'll be going all FX series Athlon 64 in the near future but if Intel doesn't get their shit into gear I could see it happening in the not too distant future, I'm hoping Intel uses it as motivation and we see a period of healthy competition but pessimism makes me think they're more likely to either completely fail fail to catch up or if they do get motivated they'll blow right past them again(I'm sure they've still got a couple dirty tricks up their sleeves as well).

It's not chip designs that are the issue -- it's the fabs to produce them. AMD was lucky to partner with TSMC, who had the breakthrough that Intel (and GloFo and Samsung and...) haven't yet made. Otherwise AMD will still be at pre-Skylake levels of performance.

Now, supposing that Intel does get their fab tech online, AMD and TSMC will have a fight on their hands, and based on the respective companies' history (that is, Intel and TSMC), it's more likely for Intel to fix their fabs than for TSMC to continue to lead.

It really doesn't seem like their process issues are a minor blip at this point. Their 10nm woes still haven't been completely worked out and now it's bleeding over into 7nm because they're scrambling to try and polish 10nm still. They still have the market position and finances to right the ship but I'm becoming less and less confident that they have the leadership and technical talent to do so. Despite the issues I have with their anti-competitive practices I actually want them to survive because they're a semi local company and I've had friends that worked for them and were treated very well but I'm starting to wonder if they're going the way of IBM or Xerox.

On the other hand TSMC seems to not be stumbling on their process node upgrades like they were a few years ago and I'm not sure why they shouldn't be expected to maintain that momentum until we hit a true wall, at this point their main issue seems to be the capacity to handle everyone wanting to use their fabs(Intel included apparently).

Well AMD kinda already did that to motherboards for high end. I didn't think I would buy an AMD board thats 370 lol. But I did. They reset that market because they have good CPUs. But I honestly don't see them raising pricing on the CPUs to intel's level. You will likely see them stick to what they have + 50 may be. But They already have increased their margins pretty damn good so I highly doubt they need to match intel with pricng. Intel has just been little greedy without competition.

I think that has more to do with the motherboard manufacturers feeling like AMD actually has something worthy of selling high end boards for. Ever since Phenom AMD had been the budget brand so even the enthusiast boards were more oriented towards the bang/buck crowd but now we're starting to see AMD boards go all out on power management features and include high end secondary chips and circuitry(LAN, Audio, USB, etc.). The only price increase that can be directly pinned on AMD is adding PCIe 4.0 to X570 since that requires more robust boards with beefier traces but you also have the option of going with an older board if you don't want PCIe 4.0.
 
It really doesn't seem like their process issues are a minor blip at this point. Their 10nm woes still haven't been completely worked out and now it's bleeding over into 7nm because they're scrambling to try and polish 10nm still. They still have the market position and finances to right the ship but I'm becoming less and less confident that they have the leadership and technical talent to do so. Despite the issues I have with their anti-competitive practices I actually want them to survive because they're a semi local company and I've had friends that worked for them and were treated very well but I'm starting to wonder if they're going the way of IBM or Xerox.

On the other hand TSMC seems to not be stumbling on their process node upgrades like they were a few years ago and I'm not sure why they shouldn't be expected to maintain that momentum until we hit a true wall, at this point their main issue seems to be the capacity to handle everyone wanting to use their fabs(Intel included apparently).
Main thing is that it's one single advancement that TSMC happened to bet right on. Intel has had issues in the past and overcome them; while it's absolutely possible that they may not, it's also highly unlikely that they'll get stuck, and even if they wind up 'a node behind', it's also clear that they still have the engineering chops to make the best of the available process as well. It's also important to point out that they're not alone here, and also that this isn't a place they'd have chosen to be stuck at given a choice. It's costing them quite a bit of earnings, and the industry has simply had to do without as there's no chance that TSMC could currently fill in where Intel falls short in pure manufacturing output.

Of what's probable, the very worst would be Intel maintaining parity with whatever TSMC can build for AMD. More likely than that is Intel sorting their process technology out, TSMC running up against the limitations that physics places on semiconductor advancements, and Intel using their engineering resources to stay a bit ahead in most metrics, something similar to Zen+ vs. Skylake perhaps.l


And I guess I should reiterate that that's not to denigrate AMD for overcoming the challenges they've faced to come up with Zen; they may not have leaped ahead, but they're still quite admirably in the game.
 
Main thing is that it's one single advancement that TSMC happened to bet right on. Intel has had issues in the past and overcome them; while it's absolutely possible that they may not, it's also highly unlikely that they'll get stuck, and even if they wind up 'a node behind', it's also clear that they still have the engineering chops to make the best of the available process as well. It's also important to point out that they're not alone here, and also that this isn't a place they'd have chosen to be stuck at given a choice. It's costing them quite a bit of earnings, and the industry has simply had to do without as there's no chance that TSMC could currently fill in where Intel falls short in pure manufacturing output.

Of what's probable, the very worst would be Intel maintaining parity with whatever TSMC can build for AMD. More likely than that is Intel sorting their process technology out, TSMC running up against the limitations that physics places on semiconductor advancements, and Intel using their engineering resources to stay a bit ahead in most metrics, something similar to Zen+ vs. Skylake perhaps.l


And I guess I should reiterate that that's not to denigrate AMD for overcoming the challenges they've faced to come up with Zen; they may not have leaped ahead, but they're still quite admirably in the game.

I don't think it's fair to characterize TSMC's position as one good bet considering how far behind they were a couple of gens back and have improved a bit more each time. I do agree that they're limited in output capacity and couldn't take up the slack even everyone wanted them to but that's a short-medium term problem if they really believe there's demand for more production capacity.

I'm still far from writing Intel off but their position is much shakier than I would have predicted possible a few years ago and the root of that is their issues with 10nm and now 7nm, they were generally way ahead of everyone else in their manufacturing nodes which made everything else easy for them. Their CPU architectures are still good and I don't doubt their capabilities to innovate there but at the same time I think AMD's approach with chiplets might be more forward thinking and if that's the case they already have a leg up on working out the associated issues.
 
I wish AMD would offer their lower core chips with this high of boost. I don't want to buy 16 cores when I barely need 8.

I am sure you have nothing to worry about. Chips with less cores will probably have higher sustained all core boost then 16 core chip.
 
The difference in most games w/smt on/off is cumulative latency. Advantage still falls toward 8 cores but I'm hopeful.
 
So it's ok for Intel to normalize the industry at said price/perf tiering for what, a solid decade and then some - and now all the sudden it's a huge red flag / issue if their competitor possibly follows suite?

I could only hope AMD does such a thing. At least we'd have more than 4c/8T / PCIe 3.0 / DDR4.
Not sure why your on the defensive and trying to say that its ok for Intel because i never said that nor is anything i typed indicate that. I genuinely hope it happens. It needs to, its been to long.
 
Given that my undervolted 3900X in an entirely inadequate motherboard regularly hits 4.75ghz on a single core and usually maintains 4.2ish all 12 cores at full load, I can absolutely see this as an attainable expectation.
 
I think AMD's approach with chiplets might be more forward thinking and if that's the case they already have a leg up on working out the associated issues.
I agree, but with the caveat that chiplets are more helpful for AMDs fabless predicament. AMD would have built - and soon will be building - monolithic dies if they could have trusted TSMC with yields and volume. The advantages for chiplets aren't really there vs. monolithic designs, it's just that in this instance AMD was able to use the chiplet approach to prove to TSMC (and everyone else) that their architecture actually had promise and was actually worth supporting. Now AMD has a partner that can help them build the CPU they've wanted to build since they first taped-out Zen.
 
I agree, but with the caveat that chiplets are more helpful for AMDs fabless predicament. AMD would have built - and soon will be building - monolithic dies if they could have trusted TSMC with yields and volume. The advantages for chiplets aren't really there vs. monolithic designs, it's just that in this instance AMD was able to use the chiplet approach to prove to TSMC (and everyone else) that their architecture actually had promise and was actually worth supporting. Now AMD has a partner that can help them build the CPU they've wanted to build since they first taped-out Zen.
It was much more than that. If you read through some old statements, it is what allowed them to come out with (to qualify) so many new CPUs in such a short time frame. They didn't have as much testing to do to push a new EPYC chip or threadripper and allowed them to compete on multiple levels with limited resources. This flexibility has allowed a top to bottom product stack that Intel was/is having a hard time keeping up with. It wasn't just about TSMC and worries of yields, while that could have had some bearing, I feel it was an additional benefit of their decision, not a driving factor.
 
Yet, the A0 revision also had an OPN.... Figure that one out. And zen2 had an OPN listed for engineering samples also. OPN is just an assigned code for the specific processor line. Not sure why you think this wouldn't be the case this time when this was the exact case last time.

Edit:.Just wanted to share link for zen2 ES with OPN...
https://www.techspot.com/amp/news/78452-amd-next-12-core-cpu-appears-benchmark-database.html

OPNs are tracking numbers for commercial products. Engineering samples cannot be ordered, so they don't have OPNs (some companies explicitly will say you that "Engineering Samples do not have an OPN.")

Non-commercial samples (ES, QS,...) have ID strings. The 2D3212BGMCWH2_37/34_N code mentioned in the techspot article isn't any OPN. It is the ID string of that engineering sample.

Same happened with Zen:

1D2801A2M88E4_32/28_N --> (first batch engineering sample 8C)

2D3151A2M88E4_35/31_N --> (second batch engineering sample 8C)

ZD3601BAM88F4_40/36_Y --> (qualification sample 8C)

R7-1800X --> (Commercial chip with OPN Tray YD180XBCM88AE; OPN PIB YD180XBCAEWOF; OPN MPK YD180XBCAEMPK)

This case is the same. The "100-000000059-14_46/37_Y" code in the OP is the ID of a Zen3 qualification sample. It isn't an OPN, and that chip isn't engineering sample.
 
Last edited:
Don't bother, he does the same thing every launch. He did it with Rome at the first leak, when the clocks were 1.4/2.2 or something similar, and not a peep when not even the 64 core parts were clocked that low, or anything close.

The misunderstandings about engineering samples and qualifications samples are traced back to first gen Zen. A pair of years ago I even wrote a post explaining the differences between QS and ES. It is evident people as you prefer to continue repeating the same mistakes again and again.

Well, he very clearly spells out he is Anti-AMD. Hard to believe anything he says about AMD...

If you were familiar with the technology, you would know what I am saying is correct and that chip mentioned in the OP is not an engineering sample. You aren't familiar and have to use ad hominems.
 
The misunderstandings about engineering samples and qualifications samples are traced back to first gen Zen. A pair of years ago I even wrote a post explaining the differences between QS and ES. It is evident people as you prefer to continue repeating the same mistakes again and again.



If you were familiar with the technology, you would know what I am saying is correct and that chip mentioned in the OP is not an engineering sample. You aren't familiar and have to use ad hominems.

Quit bullshitting. You said, concretely, that Rome Epyc would launch with the clocks that were leaked, 1.4/2.2, that they were QS and that the clocks were either already set or very close. You love to bring up your post about Zen, but conveniently leave out the bit where you were completely wrong about Zen2. I'm not putting any money on a guy who's 50/50.

Links and a quote, just so everyone knows.

https://hardforum.com/threads/new-zen-2-leak.1973013/page-8#post-1044023862

https://en.wikipedia.org/wiki/Zen_2#Server_processors


If you check the link that I gave you. I already mentioned then that the codename of a qualification sample starts with a Z whereas engineering samples start with a number. I used Ryzen samples as illustration.

The codename of the chip appeared now is ZS1406E2VJUG5_22/14_N. The decoding is

Z = Qualification sample
S = Server
140 = 1.4GHz Base
6 = Revision 6
E2 = unknown TDP
V = SP3 socket
J = 64C
U = 64x 512 KB L2 + 256 MB L3
G5 = Rome
22 = 2.2GHz Boost
14 = 1.4GHz Base

An engineering sample appeared months ago. It was 2S1404E2VJUG5_20/14_N

2 = Second gen engineering sample
S = Server
140 = 1.4GHz Base
4 = Revision 4
E2 = unknown TDP
V = SP3 socket
J = 64C
U = 64x 512 KB L2 + 256 MB L3
G5 = Rome
20 = 2.0GHz Boost
14 = 1.4GHz Base
 
Last edited:
I agree, but with the caveat that chiplets are more helpful for AMDs fabless predicament. AMD would have built - and soon will be building - monolithic dies if they could have trusted TSMC with yields and volume. The advantages for chiplets aren't really there vs. monolithic designs, it's just that in this instance AMD was able to use the chiplet approach to prove to TSMC (and everyone else) that their architecture actually had promise and was actually worth supporting. Now AMD has a partner that can help them build the CPU they've wanted to build since they first taped-out Zen.


The chiplet design was necessary for AMD o do mass-production. And yet, Ice Lake quad cores wasn't a much larger chip, and they had poor yields for the life of the chip.

It wasn't just the chiplet that made Zen 2 a massive success, it as also a decent fab. But in the world of for-hire foundry, TSMC sees the most feedback from advanced partners like Apple.

If TSMC loses the fab lead over Samsung, Apple will go back to them. So TSMC has a lot more economic reasons to keep advancing/improving the process node than Intel (and they have a lot more partners to help absorb the massive fab costs). AMD is just a secondary partner (similar in volume to Nvidia).

Apple's money will always find a new cutting-edge fab tech, so Intel is going to have to drop hundreds of billions of dollars to fix this two-process-node gap anytime soon. Intel would rather maintain their high margins, and just trickle-out new fabs (while they slowly lose sales over the next decade)

Thanks to Apple's continued need for cutting-edge, mass-produced A-seris chip revisions , Intel will always be following new tech. So AMD is definitely on the right path (wherever Apple is)
 
Last edited:
Quit bullshitting. You said, concretely, that Rome Epyc would launch with the clocks that were leaked, 1.4/2.2, that they were QS and that the clocks were either already set or very close. You love to bring up your post about Zen, but conveniently leave out the bit where you were completely wrong about Zen2. I'm not putting any money on a guy who's 50/50.

Links and a quote, just so everyone knows.

https://hardforum.com/threads/new-zen-2-leak.1973013/page-8#post-1044023862

https://en.wikipedia.org/wiki/Zen_2#Server_processors

QS usually have the finals clocks (there are some few expections) of the corresponding commercial chip.

In that old post I decoded the ID of a qualification sample of Rome. And I wrote "unknown TDP". You then ignore what I wrote and pretend that the QS corresponds to some commercial SKU in the wikipedia list. And very conveniently for your ad hominenm attack you forget your own words: "This could be a low power variant".

Besides your inability to read posts, are you still negating that the 100-000000059-14_46/37_Y chip mentioned in the OP is a qualification sample?
 
Last edited:
QS usually have the finals clocks (there are some few expections) of the corresponding commercial chip.

In that old post I decoded the ID of a qualification sample of Rome. And I wrote "unknown TDP". You then ignore what I wrote and pretend that the QS corresponds to some commercial SKU in the wikipedia list. And very conveniently for your ad hominenm attack you forget your own words: "This could be a low power variant".

Besides your inability to read posts, are you still negating that the 100-000000059-14_46/37_Y chip mentioned in the OP is a qualification sample?

The Y subfix indicates this is not an engineering sample. Moreover OPN means "Orderable Part Number" and engineering samples do not have OPN. This is the final chip or something close to it (QS or a preproduction sample).

Which one is it? QS mean's it's a final chip (or something close) or QS USUALLY are close to the final chip? You claimed in this thread that, and in the linked thread, that QS means that it's the final spec, or something incredibly close. But we have proof that's not true, and your claim was wrong. Very wrong.

My ability to read is beyond question, you're just trying to argue two different points now that I've pointed out that your claim about QS = Retail isn't always correct.

So, real easy. Does a QS mean that the clock speeds are fixed and that retail will have those same clocks, or not?

Just to clarify, I'm arguing with your claims regarding QS vs retail clock speeds, not that this isn't or is a QS

edit:typo
 
Last edited:
I dunno. I am looking to upgrade this fall/winter, and for pretty much 100% gaming usage, I am not sure why I'd buy this over a 9900k or something. By and large, clock speed still seems to be king for a lot of games.
 
I dunno. I am looking to upgrade this fall/winter, and for pretty much 100% gaming usage, I am not sure why I'd buy this over a 9900k or something. By and large, clock speed still seems to be king for a lot of games.

I'm leaning towards Zen 3 for my next overdue build, mainly cause of x.265 encoding I do (or will) for Plex. It seems I'll gain more in encoding speed than I'll lose in gaming performance. But it still feels like '6 of one, half a dozen of another' no matter which way you slice it. I either have to sacrifice and regret either my missing gaming performance, or my encoding performance. Sucks.
 
I'm leaning towards Zen 3 for my next overdue build, mainly cause of x.265 encoding I do (or will) for Plex. It seems I'll gain more in encoding speed than I'll lose in gaming performance. But it still feels like '6 of one, half a dozen of another' no matter which way you slice it. I either have to sacrifice and regret either my missing gaming performance, or my encoding performance. Sucks.

See, for me I don't even do anything encoding or really workflow related at all, this machine is just for basic productivity and gaming.
 
See, for me I don't even do anything encoding or really workflow related at all, this machine is just for basic productivity and gaming.
Guess it really depends on how much IPC increase we see.. a few days extra MHz and a decent 10% IPC increase *could* get it competing in games, and owning in everything else. It could also call short and barely close the gap ;). Nobody knows until we get some concrete performance. Judging by the difference between the 3100 and 3300x at the same speeds does give me some hope we'll see a nice increase in mid range CPUs.
 
The chiplet design was necessary for AMD o do mass-production. And yet, Ice Lake quad cores wasn't a much larger chip, and they had poor yields for the life of the chip.

It wasn't just the chiplet that made Zen 2 a massive success, it as also a decent fab. But in the world of for-hire foundry, TSMC sees the most feedback from advanced partners like Apple.

If TSMC loses the fab lead over Samsung, Apple will go back to them. So TSMC has a lot more economic reasons to keep advancing/improving the process node than Intel (and they have a lot more partners to help absorb the massive fab costs). AMD is just a secondary partner (similar in volume to Nvidia).

Apple's money will always find a new cutting-edge fab tech, so Intel is going to have to drop hundreds of billions of dollars to fix this two-process-node gap anytime soon. Intel would rather maintain their high margins, and just trickle-out new fabs (while they slowly lose sales over the next decade)

Thanks to Apple's continued need for cutting-edge, mass-produced A-seris chip revisions , Intel will always be following new tech. So AMD is definitely on the right path (wherever Apple is)
So Intel (and their customers) have no need for cutting-edge tech?

I have several bridges to sell you.
 
So Intel (and their customers) have no need for cutting-edge tech?

I have several bridges to sell you.


When you sell every processor you build,, there's less impetus to get the more advanced fab up-and-running - it's a lot less of a pain repackaging / addiing a few more cores.

Having a monopoly on x86 servers for the last decade has meant they could ride huge failures on new fab nodes after 22nm, and still continue with their massive margins. Just look at how many cores Intel has hacked together for ungodly high prices, and you'll understand why they're perfectly happy sitting on ancient process nodes.

https://www.anandtech.com/show/15955/how-to-save-6000-dollars-intel-xeon-8280-vs-6258r

If companies will pay you 10,000 dollars for a top-end product on an ancient process node, then where exactly is your cutting-edge node problem on your list of priorities?
 
Last edited:
When you sell every processor you build,, there's less impetus to get the more advanced fab up-and-running - it's a lot less of a pain repackaging / addiing a few more cores.

Having a monopoly on x86 servers for the last decade has meant they could ride huge failures on new fab nodes after 22nm, and still continue with their massive margins. Just look at how many cores Intel has hacked together for ungodly high prices, and you'll understand why they're perfectly happy sitting on ancient process nodes.

https://www.anandtech.com/show/15955/how-to-save-6000-dollars-intel-xeon-8280-vs-6258r

If companies will pay you 10,000 dollars for a top-end product on an ancient process node, then where exactly is your cutting-edge node problem on your list of priorities?
You should ask the company that has upgraded from so many nodes since the dawn of their existance.
 
You should ask the company that has upgraded from so many nodes since the dawn of their existance.


That doesn't mean they're keeping up with the rest of the universe NOW. TSMC still hs vastly higher production capacity on 7nm, AND vastly higher yields than Intel 10nm.

TSMC is already shipping their 5nm products to Apple for their iPhone launch in September/October , and their Mac chips later in the fall. Intel will not be producing 7nm chips (same process node density) until 2022.

I would call a 4 year gap between major process nodes falling behind the rest of the industry. Even though Intel's 14nm manufacturing dwarfs even TSMC, their fuckups on the 10nm node has had a aftereffect of slow expansion to other smaller fabs...because the last thing you want o do is multiply wildfires.

Do you have some reliable source that will somehow reverse this perpetual process slide?
 
Last edited:
Which one is it? QS mean's it's a final chip, or something close, or QS USUALLY are close to the final chip? You claimed in this thread that, and in the linked thread, that QS means that it's the final spec, or something incredibly close. But we have proof that's not true, and your claim was wrong. Very wrong.

My ability to read is beyond question, you're just trying to argue two different points now that I've pointed out that your claim about QS = Retail isn't always correct.

So, real easy. Does a QS mean that the clock speeds are fixed and that retail will have those same clocks, or not?

Just to clarify, I'm arguing with your claims regarding QS vs retail clock speeds, not that this isn't or is a QS

So you quote me stating that "QS usually have the finals clocks (there are some few expections)", then you ignore what I wrote, and say me that QS = Retail isn't always correct.

Besides ignoring what I am saying, your 'proof' that QSs do not always have the same clocks than commercial chips is not based in some identified QS*, but you you rely on a leak of a Rome sample with unknown TDP.

-----
* I gave one concrete example years ago. It was a six core desktop Zen, if I remember it correctly, and the difference between the QS and the commercial chip was of 200MHz or something as that.
 
Last edited:
So you quote me stating that "QS usually have the finals clocks (there are some few expections)", then you ignore what I wrote, and say me that QS = Retail isn't always correct.

Besides ignoring what I am saying, your 'proof' that QSs do not always have the same clocks than commercial chips is not based in some identified QS*, but you you rely on a leak of a Rome sample with unknown TDP.

-----
* I gave one concrete example years ago. It was a six core desktop Zen, if I remember it correctly, and the difference between the QS and the commercial chip was of 200MHz or something as that.

I'm ignoring it, because that hasn't been your point until just now.

Every time there's a leak, you pop in to explain the difference between ES and QS, and how the fact that if the leak is a QS it means the clocks are almost exactly what retail will be.

You've claimed, up until now, that QS is representative of retail.

You decoded that Rome leak as a QS.

That Rome QS, it's not a retail product. There are no Epyc Rome cpus with anywhere close to those clock speeds. You keep regurgitating that one post where you were, after the fact, able to go and show that a Zen QS had the same speeds as a retail part.

It's not like you claimed that before launch, you use it as evidence after the fact. Then try to show that it's proof going forward of how you're a reliable source.

So, in conclusion. You don't know what you're talking about, you only show up to crap up threads and then disappear. Until the next time you get an opportunity to try and BS people who haven't had the privilege of watching you twist yourself into a pretzel to avoid saying the words "I'm wrong"

Edit: To be clear, in your own words.

If you check the link that I gave you. I already mentioned then that the codename of a qualification sample starts with a Z whereas engineering samples start with a number. I used Ryzen samples as illustration.

The codename of the chip appeared now is ZS1406E2VJUG5_22/14_N. The decoding is

Z = Qualification sample
S = Server
140 = 1.4GHz Base
6 = Revision 6
E2 = unknown TDP
V = SP3 socket
J = 64C
U = 64x 512 KB L2 + 256 MB L3
G5 = Rome
22 = 2.2GHz Boost
14 = 1.4GHz Base
 
Every time there's a leak,

Not true. I have been a very long time without posting. In any case, I post what I want and when I want. Deal with it.

So, in conclusion

you misread what I write or you simply ignore what I write and replace it with your own version, which then you use to derail threads with ad hominems. You affirm that I said that Rome "would launch" with 1.4GHz clocks, and you affirm that I said "QS = Retail", but I didn't say anything of that. You are just inventing.

They are multiple explanations for that old Rome leak you keep citing: (i) the leak was fake, (ii) the QS corresponded to some LP chip still unreleased or some custom design, (iii) the chip was cancelled. Moreover, I stated that Rome QS had unknown TDP, still you insist on comparing the clocks of that QS with the clocks of commercial chips with known TDP.

All what I have done in this thread is to state that the "100-000000059-52_ 48/35 _ Y" chip mentioned in the OP is not an engineering sample as affirms Techpowerup but it is a qualification sample... and sincerely, I don't could care less about if you like my post or not.
 
Not true. I have been a very long time without posting. In any case, I post what I want and when I want. Deal with it.



you misread what I write or you simply ignore what I write and replace it with your own version, which then you use to derail threads with ad hominems. You affirm that I said that Rome "would launch" with 1.4GHz clocks, and you affirm that I said "QS = Retail", but I didn't say anything of that. You are just inventing.

They are multiple explanations for that old Rome leak you keep citing: (i) the leak was fake, (ii) the QS corresponded to some LP chip still unreleased or some custom design, (iii) the chip was cancelled. Moreover, I stated that Rome QS had unknown TDP, still you insist on comparing the clocks of that QS with the clocks of commercial chips with known TDP.

All what I have done in this thread is to state that the "100-000000059-52_ 48/35 _ Y" chip mentioned in the OP is not an engineering sample as affirms Techpowerup but it is a qualification sample... and sincerely, I don't could care less about if you like my post or not.

It is not an engineering sample.



It is a qualification sample. Clocks must be surely final. I am asking about the steeping.

Seems pretty simple to me, seemed pretty simple to you too at one point.

Keep posting, you only look like a fool with an agenda when you do. It's not like that title just dropped out of the sky.
 
All purse-swinging aside, I think AMD could eat Intel's lunch with this if the hits keep on coming the next couple of years and AMD can somehow expand capacity. Intel is floundering.
 
Lol, it says so they could in theory do it, not that there is any proof or reason to believe they will... They could release a 14-core chip too... Or 15... Or 7... Lol. They can enable or disable single cores so they could in theory (as is worded in this link) give you anything from 1 to 16 cores with 2 8 core CCDs. They could have already been doing 10 cores if they wanted, lol. This is just someone thinking out loud and calling it need.

"He predicts AMD turning up CPU core counts with this generation". So this isn't a rumor, this is someone just guessing at things.
 
To be fair, 1usmus is not your average "someone". He said he pulled it out of agesa 1081.
 
To be fair, 1usmus is not your average "someone". He said he pulled it out of agesa 1081.
Nowhere in that link/"rumor" does it say he got it from anywhere. It literally just says what I quoted above about predicting, says nothing about him finding anything or basing his prediction on anything. This would have been an important thing to include in a rumor, but this is what passes for news nowadays.

Just wanted to clarify, I do put a bit more stake in it coming from 1usmus, but thinking something might be a good idea/speculating and a rumor are 2 different things. If he has a source/reason the article completely fails to mention it.
 
Yeah, news/reporting is absolute shit these days.
Anyways, he tweeted this 3 hours ago.
 
That's sick performance. Should be able to beat Intel easily in games with those speeds.
 
  • Like
Reactions: erek
like this
Back
Top