Intel Internal Memo Reveals that even Intel is Impressed by AMD's Progress

[Neftex]

2 minutes ago, GloriousPain said:

I mean if you wanna compare their main stream platforms it's only fair. Intel didn't make above 4 cores; infact the general consensus is that the Ryzen 1700(and above obviously) is the only reason the 8700k (and sucessors) have more than 4 cores... It really was intel admitted defeat. 

im not arguing that, im glad we got ryzen

 

2 minutes ago, GloriousPain said:

I mean if you wanna look at say Skylake-E then you have to compare it to Threadripper, which AGAIN held the title... And if you're like "But intel has their Xeon with even more cores" then again look at EPYC.  

yea, and when you get up to top epyc vs top xeon platinum. the intel solutions are faster.

5 minutes ago, GloriousPain said:

It did matter, just not to gamers... But to people hosting small servers, video production, cad work, etc. it mattered greatly. Again general consensus is that it's the reason why intel quit doing Quad cores as their upper limit of the mainstream platform. 

 

There's a reason why from 2006 until 2017... Intel kept Quad core as the mainstream limit; Yet the moment Ryzen gained popularity due to it's performance crown. They upped it to 6 cores with the 8700k. And when AMD's Zen+ 2700 repeated the trend. Intel AGAIN upped their mainstream core limit to 8.  They upped their core counts TWICE in a row... After not doing it in ~10 generations. There's a reason for it, it's because they lost their crown and wanted it back. It mattered heavily.  

they factually didnt lose any performance crown when you take all options in consideration. all they lost is price/performance

 

7 minutes ago, GloriousPain said:

I mean overall Ryzen had the Price/Performance in the bag. That's obvious AMD's profit margins are anorexic in comparison to Intel's but to argue that Ryzen hasn't held any crowns or made any differences since it's release is just down right ignorant.

if threadripper and top end ryzen 1st gen stayed the same, but lower parts got significantly worse price/performance. do you think it would sell so well?

[GloriousPain]

1 minute ago, Drak3 said:

Which is blatantly wrong as Coffee Lake roadmaps have shown hexacores prior to anyone knowing anything about Ryzen outside of AMD.

I mean intel probably had rough ideas of what AMD had on the way... but yeah, my bad. Guess I shouldn't go on rants at 9am. 

[Drak3]

Just now, GloriousPain said:

I mean intel probably had rough ideas of what AMD had on the way... but yeah, my bad. Guess I shouldn't go on rants at 9am. 

It's more that general concensus is wrong, not you in particular.

[thedude4bides]

19 minutes ago, Neftex said:

you compare 4 core with 8 core in multithread ofc it looks great for the 8 core. if you want multithread then look at all the options, intel clearly had other better performing options but that performance crown didnt matter, why? ofc because of the PRICE/PERFORMANCE - my whole point

Seems like a simple enough point... price is the sole reason I'd consider AMD (because performance is "good enough" from a gaming perspective)

 

In fact, I did pull the trigger on a 2700X at one point but seller cancelled the order... which sent me back to the drawing board and I ended going with the best mainstream gaming CPU at the time which was the 8700k.

[GloriousPain]

Just now, Neftex said:

im not arguing that, im glad we got ryzen

 

yea, and when you get up to top epyc vs top xeon platinum. the intel solutions are faster.

they factually didnt lose any performance crown when you take all options in consideration. all they lost is price/performance

 

if threadripper and top end ryzen 1st gen stayed the same, but lower parts got significantly worse price/performance. do you think it would sell so well?

I'd hope so. 

 

Uh what? Not true. Epyc went upto 32 cores back in 2017... Intels just now getting to that core count. So again no matter what AMD has had the multithreaded crown albeit barely... but again it's really a stretch to even have to bring in those chips. Mainstream vs Mainstream and HEDT vs HEDT AMD won by wide wide margins in this category. 

 

I'm not sure what your 3rd question is even asking?  If they never made Zen+ or Zen2 AND made them worse? Of course if they made their things worse they'd sell worse... 

[GloriousPain]

4 minutes ago, Drak3 said:

It's more that general concensus is wrong, not you in particular.

Well that makes me feel better; but I can't find any good roadmaps about the 9000 series... Was 8cores always on the 9000 series roadmap? Because if not the point still kinda stands albeit a generation off.

[Neftex]

2 minutes ago, GloriousPain said:

Uh what? Not true. Epyc went upto 32 cores back in 2017... Intels just now getting to that core count. So again no matter what AMD has had the multithreaded crown albeit barely... but again it's really a stretch to even have to bring in those chips. Mainstream vs Mainstream and HEDT vs HEDT AMD won by wide wide margins in this category. 

just because you have 4 more cores doesnt mean they perform better...

 

3 minutes ago, GloriousPain said:

I'm not sure what your 3rd question is even asking?  If they never made Zen+ or Zen2 AND made them worse? Of course if they made their things worse they'd sell worse...  

the third question basically says again your perceived "performance crown" doesnt sell the rest of the stack

[Drak3]

11 minutes ago, GloriousPain said:

Was 8cores always on the 9000 series roadmap? Because if not the point still kinda stands albeit a generation off.

Not on road maps made for public release. Those are usually made after Intel gas dobe enough testing to see whether or not things like core count increases are worthwhile.

 

Intel's mainstream hexacore is also a part that has been delayed a few times.

[GloriousPain]

2 minutes ago, Neftex said:

just because you have 4 more cores doesnt mean they perform better...

 

the third question basically says again your perceived "performance crown" doesnt sell the rest of the stack

I mean overall in most workloads they did infact perform very similarly or marginally better. 

 

Clearly you've no idea the way businesses work. At the time I was contracted to a graphics design and video production company. All their designers were upgraded to new work stations with 1700's in them, and the more power users were given thread rippers... Why? Because they out performed their intel counterparts. 

 

Of course crowns don't matter for the parts don't have them, in which case you do an apples to apples comparison. I don't even understand what you're arguing. In multithreaded work loads; AMD's processors beat their intel equivalents. The highest end x370 chip beat the highest end z270, x470 beat z370. x399 beat x299. In terms of multithreaded workloads. 

 

1 minute ago, Drak3 said:

Not on road maps made for public release. Those are usually made after Intel gas dobe enough testing to see whether or not things like core count increases are worthwhile.

 

Intel's mainstream hexacore is also a part that has been delayed a few times.

Yeah come to think of it they probably struggled to make roadmaps given their multiple delays with 10nm and such. And how many times did intel even delay mainstream hexacore? I can't imagine they would've done it anysooner than on skylake, given until haswell-e even the ?820k part was quadcore.

[BuckGup]

I signed an NDA at my work so I can’t say exactly but when Intel says secret sauce it’s pretty saucy not gonna lie. Also Intel is in markets AMD could only dream about 

[leadeater]

8 hours ago, Neftex said:

thats 8176 not 8180 - 400mhz difference on each of the 28 cores, and you link to the FP benchmark? as you yourself pointed out there are other intel configurations than 8180 and those easily beat epyc on FP...

I know it is and an 8180 isn't 50% faster than a 8176. And no they don't easily beat EPYC on FP, but then again which FP, what data set? So pointless. EPYC is faster by 50% on more than one area, go look, I'm tired of telling  more than one and get a reply back ignoring the more than one. People that actually buy servers with these parts almost always have no interest at all in those benchmarks unless they are buying them for a render farm and only a select few, very few, businesses need that.

 

So if we could move back to something a bit more relevant like HEDT that would make a lot more sense to discuss.

[Neftex]

12 minutes ago, leadeater said:

I know it is and an 8180 isn't 50% faster than a 8176. And no they don't easily beat EPYC on FP, but then again which FP, what data set? So pointless. EPYC is faster by 50% on more than one area, go look, I'm tired of telling  more than one and get a reply back ignoring the more than one. People that actually buy servers with these parts almost always have no interest at all in those benchmarks unless they are buying them for a render farm and only a select few, very few, businesses need that.

 

So if we could move back to something a bit more relevant like HEDT that would make a lot more sense to discuss.

amd

https://www.amd.com/system/files/2018-03/AMD-SoC-Sets-World-Records-SPEC-CPU-2017.pdf

SPECRATE2017_FP_BASE=260

 

intel

https://www.spec.org/cpu2017/results/res2017q4/cpu2017-20171031-00437.pdf

SPECrate2017_fp_base = 269

 

they do easily beat epyc, just get more sockets.

 

yes people who buy servers dont only look at top performance, thats not my argument. im again saying that performance crown doesnt sell the rest of the stack. i cant believe how hard it is to understand that

[leadeater]

2 minutes ago, Neftex said:

amd

https://www.amd.com/system/files/2018-03/AMD-SoC-Sets-World-Records-SPEC-CPU-2017.pdf

SPECRATE2017_FP_BASE=260

 

intel

https://www.spec.org/cpu2017/results/res2017q4/cpu2017-20171031-00437.pdf

SPECrate2017_fp_base = 269

 

they do easily beat epyc, just get more sockets.

 

yes people who buy servers dont only look at top performance, thats not my argument. im again saying that performance crown doesnt sell the rest of the stack. i cant believe how hard it is to understand that

So you show me a single benchmark, and a quad socket configuration to show that Intel CPUs are faster but miss that you need twice as many CPUs to do it?

 

You know I have mentioned I manage hundreds of servers right? Zero 8180's here.

[Neftex]

1 minute ago, leadeater said:

So you show me a single benchmark, and a quad socket configuration to show that Intel CPUs are faster but miss that you need twice as many CPUs to do it?

 

You know I have mentioned I manage hundreds of servers right? Zero 8180's here.

you showed me FP benchmark, i told you that you can get intel config that easily beats it, you said you cant, i linked you source, you got mad, im sorry man.

 

it doesnt matter how many cpus you need to get it, if that option is available its considered.

 

idc what you have at your work. you buy what makes sense from financial, performance and probably power aspects... and thats what im saying you dont care if theres a top performer that doesnt make sense to you financially/power wise = performance crown doesnt matter

[leadeater]

12 minutes ago, Neftex said:

you showed me FP benchmark, i told you that you can get intel config that easily beats it, you said you cant, i linked you source, you got mad, im sorry man.

No you were arguing that the 8180 is faster than the 7601 and there were very few to no areas the 7601 is faster which is incorrect as there is plenty, so then you moved to arguing that you could get Intel configurations that are faster (after I raised there are better options) and probably don't understand why the performance increased nor that the server market is 80% 2 socket servers and few care about 4 sockets or more. It very much matters how many CPUs need to get, very very much because there are zero 2U4N quad socket server nodes so having to use 4 sockets removes the majority of server systems as options.

 

Also Dell has 2 socket 7601 servers scoring base 282 and 285 in the same test.

 

SPECrate 2017 Floating Point is an aggregate score of 13 different FP performance tests so you need to look at each one and see where each CPU is faster at.

 

But again if you're building a VM hosting cluster these are not the benchmarks you'd look at, neither is CPU performance alone all that important here.

 

Not only the above you brought in Xeons in to a debate about how top end consumer products help sell lesser ones, how does Xeons at all apply I have no idea.

[Neftex]

2 minutes ago, leadeater said:

No you were arguing that the 8180 is faster than the 7601 and there were very few to no areas the 7601 is faster which is incorrect as they is plenty, so then you moved to arguing that you could get Intel configurations that are faster (after I raised there are better options) and probably don't under stand why the performance increased nor that the server market is 80% 2 socket servers and few care about 4 sockets or more. It very much matters how many CPUs need to get, very very much because there are zero 2U4N quad socket server nodes so having to use 4 sockets removes the majority of server systems as options.

 

Also Dell has 2 socket 7601 servers scoring base 282 and 285 in the same test.

 

SPECrate 2017 Floating Point is an aggregate score of 13 different FP performance tests so you need to look at each one and see where each CPU is faster at.

 

But again if you're building a VM hosting cluster these are not the benchmarks you'd look at, neither is CPU performance alone all that important here.

yea and you still havent shown me the benchmarks did you.

 

you still dont get it. for each workload you can most likely get intel solution that performs better = performance crowns on intel side.

 

youre still arguing its impractical, i agree its not practical

but it doesnt change anything on the fact that there is a better performing solution.

[leadeater]

6 minutes ago, Neftex said:

yea and you still havent shown me the benchmarks did you.

 

you still dont get it. for each workload you can most likely get intel solution that performs better = performance crowns on intel side.

 

youre still arguing its impractical, i agree its not practical

but it doesnt change anything on the fact that there is a better performing solution.

We were talking about CPUs and their performance in relation to which one is better not how many sockets I need to populate before I can state I have the faster system. System performance != CPU performance. And no I didn't show them because you can easily find that information just like the CPU tests where EPYC is faster.

 

Xeons and EPYC have nothing to do with the consumer market, performance crowns, or effects on sales. How many gamers look at Xeons? How many see 7980XE overclocking videos on youtube by youtubers they watch or subscribe to? One of these two is a non zero.

[Neftex]

4 minutes ago, leadeater said:

Xeons and EPYC have nothing to do with the consumer market, performance crowns, or effects on sales. How many gamers look at Xeons? How many see 7980XE overclocking videos on youtubers by they watch or subscribe to? One of these two is a non zero.

its not so long since people were looking at xeons for gaming, u know why? again because they were good price/performance, u know the thing that sells the products

[leadeater]

3 minutes ago, Neftex said:

its not so long since people were looking at xeons for gaming, u know why? again because they were good price/performance

Yes it is long, it's ages in terms of CPU architecture cycles. LGA1366 died long ago and nobody cares anymore about that area. No gamers look at Xeons now, haven't for ages.

 

Xeons for gaming have been terrible for ages, perform much worse and cost much more. Even HEDT has been questionable for gaming since X79 and it still was then anyway.

[mr moose]

Seems somewhat appropriate for the discussion.   No matter how much evidence you have, you cannot counter an ideal not founded on evidence.

[Dylanc1500]

2 minutes ago, leadeater said:

Yes it is long, it's ages in terms of CPU architecture cycles. LGA1366 died long ago and nobody cares anymore about that area. No gamers look at Xeons now, haven't for ages.

 

Xeons for gaming have been terrible for ages, perform much worse and cost much more. Even HEDT has been questionable for gaming since X79 and it still was then anyway.

I don't know, I think my octo-socket system with Platinum 8256s, would be awesome at games.

[leadeater]

1 minute ago, Dylanc1500 said:

I don't know, I think my octo-socket system with Platinum 8256s, would be awesome at games.

*tries to put a 2080 Ti in to it*.... Why won't it fit? . Well time to pull out the PCIe external switches and a GPU chassis.

[Neftex]

10 minutes ago, leadeater said:

Yes it is long, it's ages in terms of CPU architecture cycles. LGA1366 died long ago and nobody cares anymore about that area. No gamers look at Xeons now, haven't for ages.

 

Xeons for gaming have been terrible for ages, perform much worse and cost much more. Even HEDT has been questionable for gaming since X79 and it still was then anyway.

oh yea i wonder what changed, maybe the price/performance aspect of them for gaming. oh wait...

[leadeater]

1 minute ago, Neftex said:

oh yea i wonder what changed, maybe the price/performance aspect of them for gaming. oh wait...

And the terrible CPU clocks that make them so much worse than more appropriate options.

[Dylanc1500]

11 minutes ago, leadeater said:

*tries to put a 2080 Ti in to it*.... Why won't it fit? . Well time to pull out the PCIe external switches and a GPU chassis.

Out of storage? Just order another $10,000 worth of memory and keep games in memory.