Ryzen - Return of the Jedi

[chew*]

3600C16 test done. Booted 3600 15-15-15-35 1T 1.35V same settings, will run tests.
Maybe I'm just lucky.

PS: Forgot CAS15 is rounded to 16.

Yah i booted once. Ran 32m instant crash.

That bios is running really loose so im not using it to review flare x.

[I.nfraR.ed]

Can you show MemTweakIt at 135bclk, 2666 div?
I might go back to 1002 and I will HAVE to, because of 32M, but right now testing this one. The stupid thing is I loose the saved profiles in OC menu when flashing other bios, but I know this by heart already.
I have done my previous stability tests with 2x8 and 4x8 on 1002, so I'm not too worried. Should be able to replicate these.

Here are MemtweakIt screens from my system: http://imgur.com/a/MxTEs

Maybe tWCL was slackened? Don't remember if it was lower on 1002 bios with CAS14.

[chew*]

Can you show MemTweakIt at 135bclk, 2666 div?
I might go back to 1002 and I will HAVE to, because of 32M, but right now testing this one. The stupid thing is I loose the saved profiles in OC menu when flashing other bios, but I know this by heart already.
I have done my previous stability tests with 2x8 and 4x8 on 1002, so I'm not too worried. Should be able to replicate these.

Here are MemtweakIt screens from my system: http://imgur.com/a/MxTEs

Maybe tWCL was slackened? Don't remember if it was lower on 1002 bios with CAS14.

Better yet I copied your settings. granted our cpu's are different but......either there is a huge diff in hardware or bios is super super slacked......btw these were better than my flare x in quick testing so......if these are crapping out this easy.......

[I.nfraR.ed]

Maybe IMC related, I could run the same with old bios, except I haven't tested memtest at that frequency. I had to up VDDP to 1.2V for 3600C12 to be PI32M wazza stable (1002 bios), otherwise it doesn't seem to be too hard on the memory side.

[Formula350]

Infrared: MemTweakIt doesn't show correct subtimings, if that's what you were hoping to see, at least the Tertiary ones that is. I'll double check to make sure the Secondary timings are correct.

To Chew and Infrared, where were your CPUs made? Apparently the ones made in China are better than those made in Malaysia, or at least that's what I read early on during first couple weeks of Ryzen 7's launch. The X's all seem to be China and it's the non-X that are hit-or-miss. The 'better' I think was being reported in terms of.... the IMC! Again, am going off memory, but didn't really commit much of it to long-term since my 1700X was China and so I wasn't affected by it.

Not on English Forum. I got it from the German forum, just for studying Modules. I can post the link if you will.

Yes please I only saw it mentioned on one forum, don't remember if it was German or Russian... but for some reason overclockers.de stands out in my memory.
There was only one person who mentioned it and no link was given Not really liking this 1.0.0.4a Microcode though. I can only assume it's the reason behind the memory speed issues.

[chew*]

Maybe IMC related, I could run the same with old bios, except I haven't tested memtest at that frequency. I had to up VDDP to 1.2V for 3600C12 to be PI32M wazza stable (1002 bios), otherwise it doesn't seem to be too hard on the memory side.

Problem is I have to test as the avg joe would test. here is docp loaded and I just went in edited timings to 16-16-16 saved and then bumped mem speed up.

everything else is left alone except I'm gen 1 because fury X as most users will use a real vga not an 8000 series like we would bench with.

Also avg joes probably going to be on a public bios so that eliminates betas, can't force them to use beta just to use qualified memory.

I still cant pass 32m so prime and HCI is totally out of the question. Also the xmp does jack the trc down its not the bios its the XMP.

Also Docp has the SOC covered so 1.2 soc is out of the equasion now since its setting 1.22 soc

[chew*]

Infrared: MemTweakIt doesn't show correct subtimings, if that's what you were hoping to see, at least the Tertiary ones that is. I'll double check to make sure the Secondary timings are correct.

To Chew and Infrared, where were your CPUs made? Apparently the ones made in China are better than those made in Malaysia, or at least that's what I read early on during first couple weeks of Ryzen 7's launch. The X's all seem to be China and it's the non-X that are hit-or-miss. The 'better' I think was being reported in terms of.... the IMC! Again, am going off memory, but didn't really commit much of it to long-term since my 1700X was China and so I wasn't affected by it.

Yes please I only saw it mentioned on one forum, don't remember if it was German or Russian... but for some reason overclockers.de stands out in my memory.
There was only one person who mentioned it and no link was given Not really liking this 1.0.0.4a Microcode though. I can only assume it's the reason behind the memory speed issues.

I got 2 china 2 Malaysia R5 here.....all are duds except this one. Malaysia....

1400 = 4050 stable Malaysia <-----used in 32g testing
1600 = maybe 3.5 but probably 3.4 malaysia
1500x maybe 3.5 china
1600x maybe 3.9 but most likely 3.8 china
1700 3.9 stable Malaysia <------used in 32g testing
1700x 3.8 stable maybe china
1800x 4.0 stable china

I think its more random than many want to think.......

Least you guys know AMD is not sending me cherry picked samples now lol......

[I.nfraR.ed]

I run 1.1 VSOC at 3600, VDDP 1.2, don't use DOCP or XMP, but I guess it set the tRC based on XMP profile, even though I did not load it.
The only one relatively new VGA I have is GTX980 in my main machine. But anyway, it's not advisable to run bclk that high anyway, so these 3600 tests are mostly irrelevant for the moment. We will see in May if the add more dividers and unlock subtimings.
I'm only testing 2D for now, will move to 3D soon just to see what's possible.

My CPU is China, I believe. 4GHz at 1.3V and lower on air (1.325V set in bios, but using LLC3). 4.15 at 1.46V for some multithread benches on air.
Have only ran ~30min Prime95 Blend at 4G 1.3V, so can't call it stable.

PS: 3600C14 passed memtest with 1.485V(set in bios/soft) VDIMM.

[chew*]

I run 1.1 VSOC at 3600, VDDP 1.2, don't use DOCP or XMP, but I guess it set the tRC based on XMP profile, even though I did not load it.
The only one relatively new VGA I have is GTX980 in my main machine. But anyway, it's not advisable to run bclk that high anyway, so these 3600 tests are mostly irrelevant for the moment. We will see in May if the add more dividers and unlock subtimings.
I'm only testing 2D for now, will move to 3D soon just to see what's possible.

My CPU is China, I believe. 4GHz at 1.3V and lower on air (1.325V set in bios, but using LLC3). 4.15 at 1.46V for some multithread benches on air.
Have only ran ~30min Prime95 Blend at 4G 1.3V, so can't call it stable.

PS: 3600C14 passed memtest with 1.485V(set in bios/soft) VDIMM.

I'm even less stable when I go full manual memory mode. No clue.

Some boards don't have vddp just some standby vddp to 1.050 which I tried and seemed to
do absolutely nothing.

So really even though I know for a fact it works I need to stay away from it till other boards add it I guess.

I'm not sweating it. I know a new public bios is almost done should be ready next week.

I can't complain flare x is doing advertised NP at various dividers. I think bios needs to mature before I review them though.

bclk 120 should be safe enough I have tested extensively at it but yes 135 is definitely pushing it for 24/7 use.

Pretty much anything over 120 is pushing it.

[techtrancer]

Its only a stable 100% OC stock does about 135whp with the single cam.

This does 270whp or maybe a tad more with the 2.4 turbo swap with mods

Best mod of all is insurance goes by vin # single cam ATX sure yah that's what it is

Still need to wire up the 2 step WOT box but sneakers are on now....maybe take it to the dyno if he gets me a wideband with a serial port I can plug into the laptop....

Can probably pull a few more ponies out of it but I think the head is due for an overhaul so we really just went to nail a time at the track before something goes poof......

Last leak down test showed something going on with cylinder 4 and it does not seem ring/piston/cylinder related.

You should see my neons

You should swap an B16, B18, K20 ore K24 with Turbo on it
500+ up to 1000+ is no Problem with them
But K24 does with ITB 300+ also

[chew*]

I race with them b series is slow.

K is fast but those trannies are fragile.

Itb is fun but headaches...weather changes oops back to the dyno or try to do a new street tune..

Where i live weather changes alot lol.

Ks cost to much anyway. My car is capable of 600 whp...i have $1500 into it total

[techtrancer]

Tuning is expensive
Thats normal

In Austria there is an S2000 with C32B and ITB
40000€ for swap into the S2000
But track Car!
Dayli the boy drives an lamborghini^^

[aGeoM]

@Formula350

MSI German Forum, they are not like in the English forum, people ask them for beta BIOS and if they are available a link it is posted.

Here

[Formula350]

@Formula350

MSI German Forum, they are not like in the English forum, people ask them for beta BIOS and if they are available a link it is posted.

Here

Weird that they have different policies between the forums. Or, perhaps the mods aren't as caring? haha Either way, thank you!

[chew*]

Been slammed with work plus busy squashing a bug on ocn.

Trip to microcenter today then finally can back to fooling around. Time to go cold and see how bad the imc gets cold and see what the odds are of imc coldbug with 7 chips..

My intuition tells me odds are very high all coldbug.

None of my chips appear to have the rare and elusive unicorn imc.

[chew*]

Busy busy week.

Finally almost have all the necessities gathered.



Paint brushes, conformal coating, kneaded rubber eraser, tape,towels,pens to identify OS on each HD, extra usb sticks, another 160g hd so i can make master drives and use my drive duper.

Picked up a cheap m2 drive to test with ref clock on these boards for those that are interested.

Things left to get.

30l dewar....hopefully trans am will sell me his

Threaded rod and some armaflex sheets for insulating for ln2 but for now we are all good to go with phase as soon as i prep a board

Got board prep to do and OS's to make so I am in for a very boring night zzzzzzzzzz.

[techtrancer]

Much Fun

[chew*]

Conformalled a general area front and back around socket then built a puddle control wall.

Im not worried about moisture or even drops of water. Just controlling where it can go for now. As long as it stays on my conformal area i can can give the cpu a drink of water...board will just keep working fine

Will seal cpu in socket with eraser once i am ready to install it.

[FlanK3r]

32M test pls with LN2

[chew*]

Phase change first.

I do not appear to have golden imc chips so tuning with a cb is just dumb on ln2.

I will tune on phase and pray it does not get worse on ln2 and maybe if lucky gets better.

[Buckeye]

I wonder how long these chips will run at high OC's before they start to degrade.
That was my biggest problem, workstations running high OC and after 6 months I have to start lowering the OC.
That was with Intel chips and water cooled. Tho my old QX9650 that was on Phase and did the same thing.

[chew*]

No clue bill. My 2600k is still chugging away on my m4e and that was the 5.6g chip i benched with...hard.

Anyway...time to make this a little more presentable plus im getting older my knees are killing me...

[techtrancer]

Let's do Prime under SS

Want See max clock @ 2h Prime
Also Temp scaling under SS @ max 24/7 usable Voltage

[chew*]

You dont do that under a single stage tuned for max cold...

Ss is tuned 2 ways cold temps light load or load at the cost of temps.

This one is tuned for max cold...

[ACD]

Let's see 32m