Ryzen "2" ?

DarrenM343 · 23 Apr 2018 at 07:39

AndreiD said:
Haven't noticed any change in IOPS or sequential speeds between Ryzen Balanced and High Performance, but I'm also only on SATA SSDs, might be different if you have NVMe drives. If there's an issue, I reckon it will be mitigated by the Spring Windows 10 update since they're putting the Ryzen Balanced power plan changes into the default ones.

Yep, I'm using an NVME.I'll try switching back to Performance after the next Windows update - thanks for the heads up on that.

Pants · 23 Apr 2018 at 07:59

AndreiD said:
Also, where are all of those consumer champions who were lambasting Intel for not releasing all core turbo clocks anymore? AMD is being incredibly hush-hush about all core turbo clocks for Pinnacle Ridge, where's the outrage?

Precision Boot clocks are in AMD's slide pack? https://www.anandtech.com/show/12625/amd-second-generation-ryzen-7-2700x-2700-ryzen-5-2600x-2600/5

sideways14a · 23 Apr 2018 at 08:19

Er all the boost clocks were on the press pack.

AndreiD · 23 Apr 2018 at 08:21

DarrenM343 said:
Yep, I'm using an NVME.I'll try switching back to Performance after the next Windows update - thanks for the heads up on that.

Going to get a 960 EVO soon and give it a try, hopefully the next update fixes things.
Noticed about 500 to 2000 IOPS difference when switching between Ryzen Balanced and High Performance with some more testing, but that seems to be within margin or error.

@Pants Intel published their turbo numbers too, reason why we know 8700K max 6 core turbo is 4.3Ghz. Issue with some was that Intel didn't guarantee those turbo numbers hence only marketed the maximum Turbo speed, similar to what AMD does. It's all very silly but people just wanted something to be outraged about.

AMD doesn't guarantee any turbo clocks either, no company does since they depend on workload, power envelope, cooling, and a lot of other factors.

pcfarrar · 23 Apr 2018 at 08:59

I spent all weekend gaming in VR on my new build Ryzen 2700X and Asus Prime X470 board. I didn't have a single oculus sensor disconnect or crash. Previously this was a big problem on my Z170 + i7700K system.

All my other components are the same so I guess my Z170 board was an issue previously. Either way I'm very pleased with the stability and performance.

AndreiD · 23 Apr 2018 at 09:24

Probably a motherboard issue, Oculus doesn't play nice with some ASMedia chipsets. I do have to say though, one of my favourite things about AM4 is the multitude of USB3 ports.

Loque · 23 Apr 2018 at 12:45

So after finally dialing in my ram settings and making sure it's stable, I still have cold boot issues. Not as bad as my 1700, but now im getting qcode 7A, which I can get past after a couple of resets. No idea what qcode 7A means as it's not listed in the manual or anywhere else.

DarrenM343 · 23 Apr 2018 at 12:53

AndreiD said:
Going to get a 960 EVO soon and give it a try, hopefully the next update fixes things.
Noticed about 500 to 2000 IOPS difference when switching between Ryzen Balanced and High Performance with some more testing, but that seems to be within margin or error.

I'm using a 960 Pro NVME. Didn't make copy of the stats but the IOPS changed from something like daft like 300,000 to just 50,000

, at least according to Samsung's own software. I never tried any other benchmark tool. All fine now though using Ryzen balanced.

Joeyuk · 23 Apr 2018 at 13:07

Messed about with the ram quickly. Was set at 3200 C14 which got 1804 in Cinebench. Tried 3600 C16 which got 1810 but crashed in games. Then Tried a few inbetween at C15 but all scored 3-6 points higher then 3200 so seems small gains. Did see bigger gains in RealBench and a small gain in Fire Strike. Got no idea how to tweak ram so will just leave it at 3200c14 for now as it seems pointless messing with it for small gains.

AndreiD · 23 Apr 2018 at 13:25

Over 3200 C14 you'll have better gains by lowering timings. Try and see if you can get to 3466 C14.

DarkHorizon472 · 23 Apr 2018 at 13:33

3333c14 can also be stable which is very similar to 3466 c15.

orbitalwalsh · 23 Apr 2018 at 13:33

Joeyuk said:
Messed about with the ram quickly. Was set at 3200 C14 which got 1804 in Cinebench. Tried 3600 C16 which got 1810 but crashed in games. Then Tried a few inbetween at C15 but all scored 3-6 points higher then 3200 so seems small gains. Did see bigger gains in RealBench and a small gain in Fire Strike. Got no idea how to tweak ram so will just leave it at 3200c14 for now as it seems pointless messing with it for small gains.

with gains, are you talking highest FPS or average and lowest recorded FPS? look to those more then highest FPS

Joeyuk · 23 Apr 2018 at 13:36

AndreiD said:
Over 3200 C14 you'll have better gains by lowering timings. Try and see if you can get to 3466 C14.

I'm on 3333c14 now, Cinebench showed 1-2 points and firestrike shows a small gain. I won't bother trying to max it as I mostly game so won't see many gains anyway, plus i've got no clue what i'm doing and just using the presets on the ch6.

Was a bit worried about getting the 2700x early but i've run into no problems so far like I did with my 1600. Helps that I don't need to overclock the 2700x as it works just fine at stock.

Joeyuk · 23 Apr 2018 at 13:44

orbitalwalsh said:
with gains, are you talking highest FPS or average and lowest recorded FPS? look to those more then highest FPS

tbh i didn't pay much attention. Fire strike was 300 points higher so nothing major. Assassins Creed got the same average fps but higher score overall. Just going to leave it at 3333c14 and be happy.

youdsy · 23 Apr 2018 at 13:48

How are frametimes and overall smoothness with ryzen at high refresh rates compared to a 4790k?

AndreiD · 23 Apr 2018 at 14:18

You have 1% mins in here, but your best bet is to probably wait for Zen 2 or get a Coffee Lake: https://static.techspot.com/articles-info/1613/bench/Average.png

humbug · 23 Apr 2018 at 14:26

youdsy said:
How are frametimes and overall smoothness with ryzen at high refresh rates compared to a 4790k?

From a 4790K i don't know, i went from a 4.5Ghz 4690K to what you see in my signature and its anything from the same to just about twice the performance, any stuttering i did get on the 4690K is now gone.

Having said that the 4690K didn't have SMT, yours does.

chrcoluk · 23 Apr 2018 at 14:35

sideways14a said:
Guys overclockers are a tiny minority, heck i am involved in the purchase of between 1200-1400 new base units per year for our refresh and not a single one is overclocked - most are not even capable even with the higher end K series chips.
Yes we all love to add a few hundred Mhz here, i have a 5820k doing overtime with a big water cooler but i am under no illusions i am in the very very small minority.

Its rather refreshing to see a chip that can actually do better under its own control than forcing an overclock... well unless you want to get silly cooling out - wonder what xfr would think of LN2.

these are for tech enthusiast gamers right? the 1400 units

you know the target audience of these reviews.

chrcoluk · 23 Apr 2018 at 14:39

LOLN1 said:
Anandtech reviewers may be right
Here's what happens when you follow a good guide in the bios
and activate the tasty options
I also have the video if you want i insert it on youtube

here is the problem, some say the intel chips should not be manually overclocked because its not "stock".
if they had to enable those bios options then its also not "stock".

If those options are fully stable and supported why are they disabled by default? vendor error or something else?

I would just enable the options, but I would also keep the OC intel data on the graphs, so best of both worlds.

chrcoluk · 23 Apr 2018 at 14:44

AndreiD said:
@Hotwired He specifically talks about Precision Boost Overdrive being the reason why the 2000 series chips exceed their TDP, so The Stilt didn't really miss anything. Read his thoughts on power consumption.
Issue here is that Precision Boost Overdrive, Core Performance Boost, Performance Bias, etc. are all MCE like UEFI options that AMD has now and a lot of reviewers didn't properly disable all of these 'features'. Given how much outrage came out of the MCE debacle, you'd expect there would be some for AMD's flavour.
I wouldn't even blame AMD, the fault, as with MCE, probably wholly lies with the motherboard makers trying to one up eachother.

@DragonQ There were quite a few R7 1700 samples which reviewers managed to overclock to 4.1Ghz on launch, reason why I'm reticent. Plus AMD could start saving the top samples for the Zen+ Threadripper again.
Either way if you have anything more than a 1st gen R5, it's not really worth it to 'upgrade'.

right so the reason its not enabled by default is it makes the chips exceed TDP and as such be "out of spec". So the tests with this enabled vs intel chips in "stock" config are not stock vs stock after all and now we know why the option is disabled by default, its like the MCE option in the ASUS intel bios which would put all cores at 4.7 ghz on 8700ks but the reviews were done with 8700ks at 4.4 (MCE disabled).