Soldato
Threadripper is meant to compete with Intel's x series no?
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Threadripper on Zen+ 32 Cores - Launching Q3 2018
- Thread starter Beren
- Start date
Threadripper is meant to compete with Intel's x series no?
Soldato
And that's just fine, but it's also a bit naff that you can't push things that far. Yes, though, I'll agree you can always fall back on that. You can fall back on it for core frequency, too.
Associate
- Joined
- 31 Aug 2017
- Posts
- 2,209
Well we dont yet know how far TR2 will go but from all the leaks so far it looks like 4.x GHz all core is doable.. maybe on really good water 4.3 could be seem all core which on a 32 core monster is just utterly mad.
So its OC is decent. Everyone keeps expecting 5Ghz because intel can do it with 6 cores but thats comes with plenty of hang ups like delidding ect plus of course only having a few cores to do it with.
If TR2 can do 6k plus on water in CB15 then its got some baws to it, if it can reach that with that air cooler then ... that really would be something special.
So its OC is decent. Everyone keeps expecting 5Ghz because intel can do it with 6 cores but thats comes with plenty of hang ups like delidding ect plus of course only having a few cores to do it with.
If TR2 can do 6k plus on water in CB15 then its got some baws to it, if it can reach that with that air cooler then ... that really would be something special.
Soldato
Actually as it is Intel X series is trying to compete with Threadripper, and is failing on that front.
Soldato
I certainly wouldn't expect 5GHz. For 64 threads I'd expect no more than 4.1-4.2 at best, but remains to be seen. Even 16 cores really benefit from a chiller on first gen parts. You'll be talking trying to dissipate some serious current with heavy workloads. I do agree mostly that it's not really about that, though. The more threads the better and there are people who genuinely have been dying for this amount of horsepower at a reasonable price. From an enthusiast standpoint, it's easier to pick the holes.
Soldato
Soldato
You know what i meant......
But I agree, threadripper has ripped intel a new one here. AND HERE HE IS!!!!
Welcome GavIntel to the conversation
Just pointing out the errors that seem to get thrown about here.
Associate
- Joined
- 31 Aug 2017
- Posts
- 2,209
No we wont be seeing anything like 5ghz, we will be lucky to see 4.5 XFR on 1 or 2 cores with these (that could be a pretty safe bet as the silicon is top drawer and we already know good Zen cores can do 4.5 big ones)
But if say 4.1 or 4.2 all core is there on an overclock that will be huge.
Really is daft talking about relatively cheep 64 thread single processor high speed x86 stuff not too long after 4 core parts were considered enthusiast and 6 core parts HEDT.... how the worlds changed.. and from AMD of all companies.
But if say 4.1 or 4.2 all core is there on an overclock that will be huge.
Really is daft talking about relatively cheep 64 thread single processor high speed x86 stuff not too long after 4 core parts were considered enthusiast and 6 core parts HEDT.... how the worlds changed.. and from AMD of all companies.
Soldato
It is and it isn't. I'm certainly in no rush to switch out. The number of cores and price is really the turning point here. The meme of old still reigns true.
Soldato
Didn't AMD France "leak" a performance chart recently? 2990WX clocked at 5.1GHz under LN2 giving Intel's 28-core stunt a bit of an ass-whooping?
But then again you'd expect 4 more cores and more extreme cooling to win, wouldn't you
Soldato
Depends. First of all all Threadripper chips are the top 5% best binned Ryzen+, like on last generation.
And we already know that the 2950X has advertised boost clock 4.4Ghz and that under air.
So with good motherboard, and watercooling 4.5Ghz is pretty achievable. Probably be able hit 4.6Ghz if we do proper overclocking with offsets and PBO fiddling.
True we won't see 4.5Ghz easily on the 2990WX under water, but who knows. That X399 MEG can push the power needed, and with just a dedicated 480mm rad, plus VRM cooling it could be achievable at home.
However I wouldn't expect from reviewers to do fine tune overclocking. It is too complex aspect for many of them who are used to put power and raise clocks.
Which is counter productive on the Ryzen CPUs regardless if they are AM4 or TR.
Same applies to AMD GPUs.
(ranting follows).
Trying to overclock them with brute force doesn't work either. And I could point at you all those Vega 64 reviews which are at best PATHETIC.
For example all of them boast that the Vega 64 consumes 330W+ at 1530 clock, but that is stupid and not true actually if they spent 10 minutes understand how to overclock it.
I could see only that power consumption when the GPU was overclocked to 1750 core and 1060 HBM and benching on the air and 1780 and 1090 under water.
(yes less heat means better clocks at same power consumption)
At 1630 core with 1030 HBM the card burns 276W. (Turbo mode) outperforming all the review benchmarks
At 1550 core with 945 HBM (stock boost speeds of reference card) that's down to 230W. Thats 100W+ less than the "reviewers".
At 1360 core with 800 HBM (power save) thats down to 170W.
The performance difference between Power save mode and turbo mode is 10%!!!!!! And if you bother you can get 1430 clock 900HBM at 176W by manually fiddling the card. Which makes it 5% slower than the turbo boost.
For example all of them boast that the Vega 64 consumes 330W+ at 1530 clock, but that is stupid and not true actually if they spent 10 minutes understand how to overclock it.
I could see only that power consumption when the GPU was overclocked to 1750 core and 1060 HBM and benching on the air and 1780 and 1090 under water.
(yes less heat means better clocks at same power consumption)
At 1630 core with 1030 HBM the card burns 276W. (Turbo mode) outperforming all the review benchmarks
At 1550 core with 945 HBM (stock boost speeds of reference card) that's down to 230W. Thats 100W+ less than the "reviewers".
At 1360 core with 800 HBM (power save) thats down to 170W.
The performance difference between Power save mode and turbo mode is 10%!!!!!! And if you bother you can get 1430 clock 900HBM at 176W by manually fiddling the card. Which makes it 5% slower than the turbo boost.
And is also down to cooling used. Most of the reviewers last year used normal size AIOs to benchmark the TR CPUs. Which only partially covers half of each chip.
True full size TR AIOs didn't existed.
Soldato
- Joined
- 19 Oct 2008
- Posts
- 5,952
It's hardly worth being bothered about though unless you're watching frame rate counters constantly. At 1440P and above you might as well just buy the best CPU that matches your main requirement. If someone only games, bit of browsing, on a budget then they can buy less cores (8700K or 2700). if anyone has uses that require more cores or just wants to have more cores available they're really not going to notice a difference at 1440P and above especially, unless staring at FPS counters instead of enjoying the actual game
.Not being anti-intel (quite a fan of both now when it comes to CPU) but everytime I hear a generalisation about gaming better on Intel's offerings it does make me cringe a bit, as it's dependent on a few factors and as above, whether people would notice a difference anyway.
Soldato
It makes you cringe that one part performs better than another? lol. Considering that one of the combined factors is IPC when it comes to gaming, that's pretty dumb. Turd Ripper doesn't have to be the second coming of Christ to be a decent product, but ignoring performance metrics is a step too far.
Soldato
It's hardly worth being bothered about though unless you're watching frame rate counters constantly. At 1440P and above you might as well just buy the best CPU that matches your main requirement. If someone only games, bit of browsing, on a budget then they can buy less cores (8700K or 2700). if anyone has uses that require more cores or just wants to have more cores available they're really not going to notice a difference at 1440P and above especially, unless staring at FPS counters instead of enjoying the actual game
Not being anti-intel (quite a fan of both now when it comes to CPU) but everytime I hear a generalisation about gaming better on Intel's offerings it does make me cringe a bit, as it's dependent on a few factors and as above, whether people would notice a difference anyway.
One game, total war showed 60fps vs 80fps. I'm sure most could feel that difference.
Anyway this thread is about TR. I was merely pointing out that Panos had it wrong that 1440P shows no CPU differences in gaming.
It clearly does, looking at the latest digital foundry video shows this.
I wont be replying to anymore comments about the 8700k here, this thread is not for that.
Associate
- Joined
- 31 Aug 2017
- Posts
- 2,209
Cant see 4.5 all cores but 4.5 single under XFR... yeah that could be doable.
Wonder if the bclk tweaks can also bring it up a bit like the other lesser cored chips can, people have been seeing 4.4 and 4.5 Ghz 2700s for ages now with good cooling and the right options in the bios.
Wonder if the bclk tweaks can also bring it up a bit like the other lesser cored chips can, people have been seeing 4.4 and 4.5 Ghz 2700s for ages now with good cooling and the right options in the bios.
Soldato
The point is which total war? It does matter because is down to the game also.
Total War Warhammer 1 (Rome 2, Attila also) was forcing itself to run on the last thread of a given CPU.
If you had an i5 or turn off the SMT on the Ryzen the performance was improved.
Total War Warhammer 2, was forcing itself to run the "main thread" to last true core, in addition it utilised the first two cores on a CPU.
It stayed clear of running against threads because of the performance issues.
Also don't forget Ryzen 1 was initially plagued (until Dec 2017) with Windows 10 scheduler issues, which are now patched.
Finally in relation to continuing your failed argument. Have look at reviews here. Check all games against all CPUs at 1440p
https://www.guru3d.com/articles_pages/amd_ryzen_7_2700_review,20.html
Also check here various motherboards with the same 2700
https://www.guru3d.com/articles_pages/gigabyte_b450_aorus_pro_review,14.html
Permabanned
They have, I had mine boosting near 4.5, however the voltage was quite a lot and I wasn't sure it was especially safe.
Going to be interesting to see how these new Threadrippers turn out.
Soldato
Using a GTX 1080 SMH
Give it a proper GPU and see how it goes. Better yet wait for the 1180 and see the gap increase even further.
Why are we even discussing gaming performance in a TR thread?? Nobody is buying one of these to game on LOL.
Soldato
Using a GTX 1080 SMH
Give it a proper GPU and see how it goes. Better yet wait for the 1180 and see the gap increase even further.
Why are we even discussing gaming performance in a TR thread?? Nobody is buying one of these to game on LOL.
35 game benchmarks with GTX1080Ti
https://www.techspot.com/review/1655-core-i7-8700k-vs-ryzen-7-2700x/
Enjoy as it includes the ones you posted above.
Soldato
Did you actually read the review????
It clearly shows the 8700k is faster even at 1440P. I cannot wait for the 11 series of GPU's to arrive and put this argument to bed.
This whole 8700k is pointless above 1080P argument needs to stop there are 20fps differences is some titles, that can be a GPU upgrade to some.
Again, nobody is gaming on TR.
I'm not even sure why we are discussing this in here.