*** AMD ThreadRipper ***

[TNA]

Fixed it for ya

 

[marms]

Yup, for the OEMs intel has a staggering i9-9980HK 45W mobile chip running at 2.4GHz/5.0GHz and with iGPU UHD Graphics 630 https://www.intel.com/content/www/us/en/products/processors/core/i9-processors/i9-9980hk.html

What can AMD offer for those millions of office PCs?

Nothing!

At $583 each i don't think you'll see many office PCs with 9980hk processors, companies buying large numbers of oem office computers want an entire computer for that price. And you said the 3900x was overpriced lmao.

 

[4K8KW10]

AMD Ryzen 9 3950X 16 Core PassMark Benchmark Shows It’s Faster Than Intel’s 28 Core Xeon-3175X

Another AMD Ryzen 9 3950X CPU Benchmark Leaks Out & It Shows Why Intel Should Be Rethinking Their HEDT Strategy
https://wccftech.com/amd-ryzen-9-3950x-16-core-cpu-passmark-benchmark-leak/

 

[Deleted member 209350]

AMD Ryzen 9 3950X 16 Core PassMark Benchmark Shows It’s Faster Than Intel’s 28 Core Xeon-3175X

Another AMD Ryzen 9 3950X CPU Benchmark Leaks Out & It Shows Why Intel Should Be Rethinking Their HEDT Strategy
https://wccftech.com/amd-ryzen-9-3950x-16-core-cpu-passmark-benchmark-leak/

Yikes, intel are miles behind in the HEDT department. How can a 16c be dominating a 28c like that?

 

[LePhuronn]

How can a 16c be dominating a 28c like that?

Exactly the same way a dual-core Haswell i3 dominated a 5GHz 8 core Piledriver FX: ****-poor architecture.

 

[4K8KW10]

Yikes, intel are miles behind in the HEDT department. How can a 16c be dominating a 28c like that?

This particular benchmark is the best case for AMD cores - it heavily favours them.

 

[Deleted member 209350]

This particular benchmark is the best case for AMD cores - it heavily favours them.

In what way does it heavily favour them? (genuinely asking)

 

[Deleted member 209350]

Exactly the same way a dual-core Haswell i3 dominated a 5GHz 8 core Piledriver FX: ****-poor architecture.

Must be very poor architecture, but for something thats only a year old and is as powerful as it looks on paper (28c SMT) its not good at all... Cannot wait to see what kind of numbers the threadripper chips are going to put up! Probably going to surpass even the epyc numbers

 

[Minstadave]

Yikes, intel are miles behind in the HEDT department. How can a 16c be dominating a 28c like that?

Clock speeds and heat issues are Intels problem. The Xeon is a 255W part and boosts to just 3.8ghz.

The 3950X is 105W, can boost higher, has higher base clocks and thermals arent a major issue. It's also quarter of the price (admittedly the Xeon is an enterpise chip).

AMDs 7nm headstart and chiplet design is really starting to show promise.

 

[4K8KW10]

In what way does it heavily favour them? (genuinely asking)

In a similar way as to how SuperPi heavily favours intel cores.
Only the developers know best.

https://www.techpowerup.com/review/intel-core-i9-9900ks/4.html



https://www.cpubenchmark.net/cpu.php?cpu=AMD+Ryzen+7+3700X&id=3485 Ryzen 7 3700X = 23860 points
https://www.cpubenchmark.net/cpu.php?cpu=Intel+Core+i9-9900K+@+3.60GHz&id=3334 Core i9-9900K = 20192 points

 

[Deleted member 209350]

Clock speeds and heat issues are Intels problem. The Xeon is a 255W part and boosts to just 3.8ghz.

The 3950X is 105W, can boost higher, has higher base clocks and thermals arent a major issue. It's also quarter of the price (admittedly the Xeon is an enterpise chip).

AMDs 7nm headstart and chiplet design is really starting to show promise.

For sure,

Im actually interesting to see what intel can offer once they've ditched the 14nm. They've taken a real beating this year, so hopefully this spurs them on to do well next year. Much like how Apple were truly awful for the past couple years, and as a result of their stock taking a massive dip, they have actually tried again this year and made good improvements in a lot of areas. I feel Intel are in the same boat and 2020 will be a good year for them.

Competition is good for the consumers, so people shouldn't want Intel to fail (though them taking an ass kicking in the short term is nice xD)

 

[4K8KW10]

Competition is good for the consumers, so people shouldn't want Intel to fail (though them taking an ass kicking in the short term is nice xD)

The only way we see competition is if intel fails. Because they occupy 90-95% of the servers/OEMs market and with dodgy practices while they still have some remaining competitiveness.

 

[Deleted member 209350]

The only way we see competition is if intel fails. Because they occupy 90-95% of the servers/OEMs market and with dodgy practices while they still have some remaining competitiveness.

Well obviously. Up until Epyc/Ryzen released, there was literally no competition, and as a result Intel definitely got very very lazy, and because of that they have actually fallen behind and are now scrambling to get upto date.

In an ideal situation, you would have both AMD and intel both upto date on technology so we would have a good enough choice of picking what we want.

The way I see things going, I think Intel is going to the better option for single core performance with their clock speeds, and AMD will be the better choice for multithreaded workloads.

 

[LePhuronn]

I think Intel is going to the better option for single core performance with their clock speeds

Except for the fact that all 10nm parts so far have a clock speed regression. There is a big IPC uplift coming to compensate, but Intel will have lost their clock speed advantage, and if AMD match or suprpass Intel's IPC at the same (or possibly) higher clocks, then Intel have absolutely zero.

 

[Deleted member 209350]

Except for the fact that all 10nm parts so far have a clock speed regression. There is a big IPC uplift coming to compensate, but Intel will have lost their clock speed advantage, and if AMD match or suprpass Intel's IPC at the same (or possibly) higher clocks, then Intel have absolutely zero.

The only thing intel have right now is the clock speed advantage, if they don't have that then they are really going to be screwed, because its the only advantage they have over AMD atm

 

[humbug]

In a similar way as to how SuperPi heavily favours intel cores.
Only the developers know best.

https://www.techpowerup.com/review/intel-core-i9-9900ks/4.html



https://www.cpubenchmark.net/cpu.php?cpu=AMD+Ryzen+7+3700X&id=3485 Ryzen 7 3700X = 23860 points
https://www.cpubenchmark.net/cpu.php?cpu=Intel+Core+i9-9900K+@+3.60GHz&id=3334 Core i9-9900K = 20192 points

Is this one of Intel's "Real world performance" benchmarks?

SuperPI uses the completely '1995' X87 extension, which Ryzen doesn't have, because its useless, it runs in software emulation mode on Ryzen which will choke its performance.

 

[Emlyn_Dewar]

Edit: double post somehow

 

[rtho782]

Do we know when the review embargo lifts on the 3960?

They are supposed to be released in a week, the same date as 3950.

 

[Deleted member 209350]

Do we know when the review embargo lifts on the 3960?

They are supposed to be released in a week, the same date as 3950.

I thought the 3950 was already released? I have seen plenty of reviews of them all over youtube, so the the 3960 must have a later date

 

[humbug]

25/11/19 for the 3950X

And it seems for Threadripper 3 too...

https://www.cgdirector.com/threadripper-3000-platform-details-release-date/

Update: AMD Announces Threadripper 3000 CPUs
Threadripper 3000 CPUs will be available to purchase on 25th November 2019. AMD has announced some details regarding these new processors and I thought it warranted an update here.