Soldato
- Joined
- 7 Dec 2015
- Posts
- 3,043
3400C14 DIMM modules are about £160 more expensive.
-
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.
Intel to launch 6 core Coffee Lake-S CPUs & Z370 chipset 5 October 2017
- Thread starter AthlonXP1800
- Start date
3400C14 DIMM modules are about £160 more expensive.
B Die memory is about £20 more than 2400Mhz SK Hynix.
The stock HSF has been fine for me.
Any UK price for the 8400?
Feel sorry for the countries with high VAT.
Current Swiss prices
7350K £125
7600K £172
7700K £229
Ryzen 1700 is £237 for comparison.
But when the 8th Gen come along, I'm expecting a bigger gap to the 7th Gen, something similar to UK pricing, 7700k is very tempting now since my Flight Sims love Ghz but think it will have to be the 8700k!
Current Swiss prices
7350K £125
7600K £172
7700K £229
Ryzen 1700 is £237 for comparison.
But when the 8th Gen come along, I'm expecting a bigger gap to the 7th Gen, something similar to UK pricing, 7700k is very tempting now since my Flight Sims love Ghz but think it will have to be the 8700k!
Google search. And find overclocks using P-states, not those who put 1.4v on the CPU and let it run.
Soldato
- Joined
- 16 May 2007
- Posts
- 3,221
Soldato
- Joined
- 16 May 2007
- Posts
- 3,221
The stock wraith spire cooler for ryzen is very good and easily copes with 3.7 at low temps. I would not trust it at 3.9.
Intel coolers in comparison are very poor or not even provided.
Waiting now for prices and proper reviews, my guess is the 1600 will be much better value for money unless you want to pay a premium for a relatively small performance increase.
Cache sizes and memory bandwidth does differ. Its why 1200/1300X are behind.
Also 1 point here or there isnt that big of a margin. Might ve even smaller than 1 point without rounding
I've not done it on an R7 but have on a good few R5 1600's, some of them are stubborn with voltage, others much more friendly. Stock at 3.925GHz at 1.3575v (approx) with temps in the high 70's or 80's in AIDA, and gaming benchmarks like Heaven drops to high 60's, that with just a single 120mm push/pull setup up in the case.
Have to say, I'm not saying it's the best solution but it's more than viable even more so if you have a board that does pstate overclocking.
Caporegime
- Joined
- 18 Sep 2009
- Posts
- 30,296
- Location
- Dormanstown.
It's the ryzen 3 I'm on about, a pretty big difference. But fair point. I was drunk anyway.
Never seen any charts that showed Ryzen IPC that high before.
I've a feeling if overclocking doesn't get any better on Ryzen, then the actual competitor is going to be the i5-8400 (non-K) at £175, which is full 6 cores, no HT but a boost clock of 3.8GHz on all cores, and with the slight IPC advantage is going to be as fast as an R5 at 4.0GHz+ also it's a nice 65w part so even with the dire thermal toothpaste in the chip hopefully won't run as hot as hell.
If one of the board vendors pulls a miracle of Bclk overclocking on the Z370, like they did with the Z170 (before Intel blocked it) then it's going to be a really hard choice. Dead socket, but could last you 3-5 years if that was the case... it's fun time for the enthusiast.
I think the surprising thing to see (although I kind of knew) is how little IPC Intel have gained since moving onto 14nm with Broadwell.
Adsitional instruction sets mean other programs may see more, but on a like for like basis not much for years now.
It's pretty inane to begin with to rely on 1 workload when testing for IPC, I'm not sure what guru3d is even trying to do there. We need something like these Anandtech articles, various different types of workloads:
http://www.anandtech.com/show/11550...core-i9-7900x-i7-7820x-and-i7-7800x-tested/16
http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/9
I'm not sure why Anandtech haven't done one with Ryzen/TR yet, they usually do it for 'new' Intel architectures. Maybe they'll do one when they get some Coffee Lake chips.
I'd really like to see one for Ryzen chips, I can imagine it will be on par with Skylake/Kaby Lake for rendering, somewhere around Ivy for encode and probably Haswell for the rest.
Edit: Forgot PCPer did a limited clock for clock comparison:
https://www.pcper.com/reviews/Proce...d-Zen/Clock-Clock-Ryzen-Broadwell-E-Kaby-Lake
http://www.anandtech.com/show/11550...core-i9-7900x-i7-7820x-and-i7-7800x-tested/16
http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/9
I'm not sure why Anandtech haven't done one with Ryzen/TR yet, they usually do it for 'new' Intel architectures. Maybe they'll do one when they get some Coffee Lake chips.
I'd really like to see one for Ryzen chips, I can imagine it will be on par with Skylake/Kaby Lake for rendering, somewhere around Ivy for encode and probably Haswell for the rest.
Edit: Forgot PCPer did a limited clock for clock comparison:
https://www.pcper.com/reviews/Proce...d-Zen/Clock-Clock-Ryzen-Broadwell-E-Kaby-Lake
Soldato
- Joined
- 16 May 2007
- Posts
- 3,221
I've a feeling if overclocking doesn't get any better on Ryzen, then the actual competitor is going to be the i5-8400 (non-K) at £175, which is full 6 cores, no HT but a boost clock of 3.8GHz on all cores, and with the slight IPC advantage is going to be as fast as an R5 at 4.0GHz+ also it's a nice 65w part so even with the dire thermal toothpaste in the chip hopefully won't run as hot as hell.
If one of the board vendors pulls a miracle of Bclk overclocking on the Z370, like they did with the Z170 (before Intel blocked it) then it's going to be a really hard choice. Dead socket, but could last you 3-5 years if that was the case... it's fun time for the enthusiast.
The choice at all price and performance levels is now very good, before AMD stepped back in Intel was following a similar approach to Apple , premium prices and minor updates for each generation.
Imagine the obscene margins they have on the 7700K, a ~123mm2 die that's selling around the £320 mark, I wouldn't be surprised if they're around 80%.
The issue there is not every chip is the same.
My 1700 at 1.35v on a stock cooler is a toasty 70-80 under AIDA, handbrake with AVX will be a lot higher.
Pstates doesnt mean **** for anything other than idle temps. Even then the difference is negligible versus not using it.
The stock cooler is good, but it isn't something I'd recommend overclocking with.
Do you have a ryzen CPU?
We were discussing about single core IPC, which Guru3d is testing PROPERLY. All CPUs at same speed, and see how they scale on single core test.
There bound to be 1-2 points difference between various ram speeds, motherboards, and topologies. But that isn't breaking factor, is as basic as it can be.
You post tests that are relative and depending to multiple parameters. Ram speed, CPU speed, motherboard, core count, potential software optimisations even cooling.
With cherry on top, FOUR (4) architectural different system, where performance matters on multi thread multi core benchmarks.
Infinity Fabric (AMD Ryzen - Ram speed), Mesh (SkylakeX - Ram Speed), Ring with speed running at CPU speed (Kabylake/X, Skylake), Ring with speed running at different speed than core (Broadwel-E, Haswel-E - uncore speed).
@Panos I don't think you understand what IPC is if you think 1 workload is all that's needed to determine IPC differences between various architectures.
It's actually a very inane test when taken on its own, because you could just as well test a heavily vectorized workload and then Ryzen will look like something from 2007 compared to Skylake-X.
I don't think you even read the articles I linked if that's the conclusion you came to, and you can use some of your nitpicking to completely discount the inane guru3rd ""IPC"" test.
It's actually a very inane test when taken on its own, because you could just as well test a heavily vectorized workload and then Ryzen will look like something from 2007 compared to Skylake-X.
I don't think you even read the articles I linked if that's the conclusion you came to, and you can use some of your nitpicking to completely discount the inane guru3rd ""IPC"" test.
Properly testing is using 1 test?
Have a read of this proper testing https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/
Before you cry any foul play or intel shill here, take a look at who the author of that article is first
