Core 9000 series

drunkenmaster · 9 Oct 2018 at 11:32

Couple things to point out when looking at Intel's numbers.

Firstly they state that the performance may not reflect the performance with all security patches installed/enabled, which implies that the performance shown is best case and that maybe even some of the 'mitigations' they promise with this gen which are actually only microcode/software fixes from launch, aren't enabled.

Second..

http://www.principledtechnologies.com/Intel/PC_gaming_processor_study_interim_1018.pdf

If you look near the bottom, one of the reasons for some big differences and the insane 50% higher gaming performance than a 2700x... is because they likely disgustingly, disabled 4 cores on the 2700x. They enabled game mode on ALL AMD systems. On threadripper that is somewhat understandable as it can help, on a 2700x it disables 4 cores. So they are benching a 2700x in 4 core mode against a 8 core Intel chip and claiming a massive advantage.

Also when you see the Geil vs Corsair benching a couple pages back, Intel had XMP settings enabled, AMD did not, they just set it to the speed supported by the platform. It's fairly surprising that they used the stock supported memory speed for AMD and Intel already in that AMD supports higher memory speed.

So they've pretty much crippled 2700x performance and even Intel performance may not be what you get from a stock 9th gen system when you install it and find all microcode and security software enabled and lower performance.

https://youtu.be/6bD9EgyKYkU?t=440

Guy on youtube who was covering this, I set it to start at the benchmark results that show the difference between the shown benchmark results and real world. You can watch the earlier part but he mostly covers what I said, XMP timing difference and questioning why the Ashes result stands out as wrong to him.

muon · 9 Oct 2018 at 11:35

Yeh, hardwareunboxed covered this.

Intel got XMP settings whilst AMD didn't...

Dirk Diggler · 9 Oct 2018 at 11:40

@drunkenmaster wait for the chips to get into normal hands and see reviews, there's far too many shills these days.

Panos · 9 Oct 2018 at 11:40

muon said:
Yeh, hardwareunboxed covered this.

Intel got XMP settings whilst AMD didn't...

And also the 2700X run as a quad core.

Screw Intel. 2920X here I come...

icehot666 · 9 Oct 2018 at 11:41

drunkenmaster said:
Couple things to point out when looking at Intel's numbers.

Firstly they state that the performance may not reflect the performance with all security patches installed/enabled, which implies that the performance shown is best case and that maybe even some of the 'mitigations' they promise with this gen which are actually only microcode/software fixes from launch, aren't enabled.

Second..

http://www.principledtechnologies.com/Intel/PC_gaming_processor_study_interim_1018.pdf

If you look near the bottom, one of the reasons for some big differences and the insane 50% higher gaming performance than a 2700x... is because they likely disgustingly, disabled 4 cores on the 2700x. They enabled game mode on ALL AMD systems. On threadripper that is somewhat understandable as it can help, on a 2700x it disables 4 cores. So they are benching a 2700x in 4 core mode against a 8 core Intel chip and claiming a massive advantage.

Also when you see the Geil vs Corsair benching a couple pages back, Intel had XMP settings enabled, AMD did not, they just set it to the speed supported by the platform. It's fairly surprising that they used the stock supported memory speed for AMD and Intel already in that AMD supports higher memory speed.

So they've pretty much crippled 2700x performance and even Intel performance may not be what you get from a stock 9th gen system when you install it and find all microcode and security software enabled and lower performance.

https://youtu.be/6bD9EgyKYkU?t=440

Guy on youtube who was covering this, I set it to start at the benchmark results that show the difference between the shown benchmark results and real world. You can watch the earlier part but he mostly covers what I said, XMP timing difference and questioning why the Ashes result stands out as wrong to him.

Yeah and if this is the case, it's even less tempting at current prices, double the price for at best 20% performance increase.

awaybreaktoday · 9 Oct 2018 at 11:44

2800X 4.5GHz base, 5.1Ghz boost, 7nm incoming...

muon · 9 Oct 2018 at 11:45

Panos said:
And also the 2700X run as a quad core.

Screw Intel. 2920X here I come...

Don't think Steve mentioned that. Wow, that is terrible, just seen it in the pdf. You sure that disables cores rather than HT? (whatever happens, only supposed to be used on threadripper https://download.amd.com/documents/ryzen-master-quick-reference-guide.pdf)

awaybreaktoday said:
2800X 4.5GHz base, 5.1Ghz boost, 7nm incoming...

7nm Zen+ hasn't been rumoured at all.

Jim Jervis · 9 Oct 2018 at 11:48

£500 for a 9700k / £1,100 for a 2080ti, consoles here I come.

Zhuni · 9 Oct 2018 at 11:51

Was ready for new build but will carry on happily with my one X until things change.

Pc market is eating itself

icehot666 · 9 Oct 2018 at 11:51

Yeah just looked into it, my single thread 2600k performance using cpu-z benchmark scores 439 - ryzen 2700x only scores 466, 8 years later and that's the best increase we've got?? 8700k is 530, so I'll be interested to see what the 9900k is.

RavenXXX2 · 9 Oct 2018 at 11:53

Lol cpuz bench, is that what you base overall performance on. The 2700x crushes sandy bridge.

icehot666 · 9 Oct 2018 at 11:55

RavenXXX2 said:
Lol cpuz bench, is that what you base overall performance on. The 2700x crushes sandy bridge.

It's just a quick idea to compare my actual CPU to others. Besides it crushes it in multi-thread performance, not single thread.

And what it says to me, is we've basically gotten nowhere in 8 years, other than adding more cores on top of more cores.

andicole0 · 9 Oct 2018 at 11:56

Don't bother with all those performance stats, all the info for a considered choice is here - https://www.pcworld.com/article/330...k-vs-ryzen-2800x-totally-fake-benchmarks.html
Andi.

Journey · 9 Oct 2018 at 12:02

Jim Jervis said:
£500 for a 9700k / £1,100 for a 2080ti, consoles here I come.

Why? You don't have to buy the very top end components to game well on a PC as I am sure you already know. Heck you can buy an R5 2600 build with an RX580 for a shade over £600, and happily play at 1440p which is better than you can do with a console.

If you become obsessed with owning the best, of the best, of the best then you are gonna get ripped off, no matter what the hobby, if it's not a hobby and just something to entertain you then there's no need to even be looking at a CPU above £160.

ZXSpectrum · 9 Oct 2018 at 12:03

icehot666 said:
It's just a quick idea to compare my actual CPU to others. Besides it crushes it in multi-thread performance, not single thread.

And what it says to me, is we've basically gotten nowhere in 8 years, other than adding more cores on top of more cores.

That's not the case at all. The only people that will take this information and see it how you have are people that use software that do not utilise all the cores. All software now is being modified to take into consideration multicore cpus. While, yes, there have been small advances in single threaded performance the multithreaded architecture is paramount in todays pcs. Otherwise we would all be using dual core i3.

SiDeards73 · 9 Oct 2018 at 12:06

awaybreaktoday said:
2800X 4.5GHz base, 5.1Ghz boost, 7nm incoming...

Pass whatever it is your smoking homie, i want some of that

icehot666 · 9 Oct 2018 at 12:10

ZXSpectrum said:
That's not the case at all. The only people that will take this information and see it how you have are people that use software that do not utilise all the cores. All software now is being modified to take into consideration multicore cpus. While, yes, there have been small advances in single threaded performance the multithreaded architecture is paramount in todays pcs. Otherwise we would all be using dual core i3.

I didn't say multicore wasn't important, I just want multi-thread AND single thread improvements - multicore improvements to me seem the lazy way out, yes it's important, but the difference say in an 8700k to a 2600k apart from slight IPC improvements, are just 2 more cores, and a clock increase - well I've already done the clock increase myself. You can see it in most benchmarks - you divide the results up on a per core basis, and then on a per clock basis, and things even out very quickly between all of them. Why? Because there's not been very much gain in IPC single thread performance.

CAT-THE-FIFTH · 9 Oct 2018 at 12:15

drunkenmaster said:
Couple things to point out when looking at Intel's numbers.

Firstly they state that the performance may not reflect the performance with all security patches installed/enabled, which implies that the performance shown is best case and that maybe even some of the 'mitigations' they promise with this gen which are actually only microcode/software fixes from launch, aren't enabled.

Second..

http://www.principledtechnologies.com/Intel/PC_gaming_processor_study_interim_1018.pdf

If you look near the bottom, one of the reasons for some big differences and the insane 50% higher gaming performance than a 2700x... is because they likely disgustingly, disabled 4 cores on the 2700x. They enabled game mode on ALL AMD systems. On threadripper that is somewhat understandable as it can help, on a 2700x it disables 4 cores. So they are benching a 2700x in 4 core mode against a 8 core Intel chip and claiming a massive advantage.

Also when you see the Geil vs Corsair benching a couple pages back, Intel had XMP settings enabled, AMD did not, they just set it to the speed supported by the platform. It's fairly surprising that they used the stock supported memory speed for AMD and Intel already in that AMD supports higher memory speed.

So they've pretty much crippled 2700x performance and even Intel performance may not be what you get from a stock 9th gen system when you install it and find all microcode and security software enabled and lower performance.

https://youtu.be/6bD9EgyKYkU?t=440

Guy on youtube who was covering this, I set it to start at the benchmark results that show the difference between the shown benchmark results and real world. You can watch the earlier part but he mostly covers what I said, XMP timing difference and questioning why the Ashes result stands out as wrong to him.

DM,I am surprised you have forgotten about the company doing the benchmarking for Intel!!

Intel used a company called Principled Technologies to do the benchmarks which sounded familar.

Then I realised 6 years ago I saw this when it was talking about the use of Intel using a certain benchmark suite from them to show X86 being better than ARM with regards to power effiency:

http://www.xtremesystems.org/forums...y-in-Tablets&p=5150663&viewfull=1#post5150663
https://www.prweb.com/releases/hdxprt/11-02-2010/prweb4699694.htm

They are closely aligned with Intel,and then look at the Intel chap commenting in that article.

Now look at BAPCo who were behind Sysmark and was founded by the same Intel employee mentioned in those links(who is VP, Sales & Marketing Group & GM, Platform Evaluation & Competitive Assessment):

https://www.businesswire.com/news/h.../BAPCo-Announces-EECoMark-TM----Joint-Venture
https://www.crunchbase.com/person/shervin-kheradpir#/entity
https://www.theregister.co.uk/2000/07/24/intel_and_bapco_just_good/

jigger · 9 Oct 2018 at 12:21

IT Troll said:
Still faster though. Happy days.

Faster than what and in what? Assuming you are talking gaming, yeah it's likely to have an edge over most other chips but maybe not all, the problem is it will be barely noticeable and certainly not worth the money.

awaybreaktoday · 9 Oct 2018 at 12:33

SiDeards73 said:
Pass whatever it is your smoking homie, i want some of that

UK Retailers Gullable Muppets drug.... Its really powerful stuff