Overclock.net banner

4721 - 4740 of 12401 Posts

·
Registered
Joined
·
49 Posts
Tested my dual rank kit some more last night, tried for 3466 14-14-14-28-42 272 1.41v DRAM 1.0v SoC which booted fine but was getting errors really fast in memtest, more than likely I needed more dram voltage, but 3400 with the exact same settings is about 1100% memtest stable which is cool. I ran a quick and dirty test at 3400 with 1.4v vdimm instead of 1.41 and it threw one error out of 10 tests @150%, so that little bump in voltage seems to have helped. Might try 3466/3500 at 1.43-45v or so next and see if that's possible.
 

Attachments

·
DRAM Obsessor
Joined
·
1,033 Posts
4.57 Ghz????? Is this a 2700x? You've got a diamond chip cooled under chilled water or what lol.
Currently i am running 102.4 bclk sticking to High performance power plan because 4.45ghz in single core give me crashes, other than that DPC is lower on high perf plan and less microstutters in games.
Basically i run 4.35ghz all the time unless im really heavy loading all cores, llc 4 + PE level 2
ye hes using hacks check hes temperatures.... Everyone knows that AMD's cpud love to be sub 70c

Not chilled water, well I guess you could say it is. I built a shelf up at the window sill above my desk and the computer is pulling in (11c currently) outside air from the side and front fans cooling the H115i pro AIO and all components in the case. It will get down to 3c tonight but the air usually is about 5c warmer than the outside air once it hits the temp sensor at the fan intake. I have also sealed up my case with weather stripping and sealed all unused intake areas as not to get warm air from my house mixing with the intake air and I just have it exit the top of the case into my room. I remain warm while my comp doesnt lol
 

·
Registered
Joined
·
462 Posts
3600 CL14 15 14 14 28 42 270 Ram 1.42v CPU 4.1GHz Offset + 0.01250v SoC 1.00625v - Preliminary bench tests - CB11 Perf Bias.

Very impressive!!! However too much freezing going on to use this Bias setting. Testing continues.
very nice dude well done
 

·
Computer Enthusiast
Joined
·
2,887 Posts
very nice dude well done
Thank you! Still testing.... OC is holding after power off, though there have been a couple of reboots where memtest only got to about 500%, but that was with other stuff running - I just wanted to see how robust the OC is. I have managed to boot to desktop @ 3666MTs a few times, but looks like 3600 is the fastest I can go at with CL14. We will see :)

I may relax my CPU to 4GHz, which may give me a little more Ram OC ability, we'll see, but will certainly bring down power requirements - and temps would be better too.
 

·
Banned
Joined
·
403 Posts
doing great but yeah it is a heat warmer. I think 4.3ghz is about tops. But you have more to play around with in there in the bios. Probably don't matter a whole lot really. Super Impressed.
 

·
Registered
Joined
·
138 Posts
Update on the 3605MHz with 1103.

Previous settings still needed tweaking, especially for CPU, no longer have high temps and freezes due to CPU.

Boosts to almost 4.5GHz, screenshot shows 4470MHz on two cores on Balanced Power Plan.

Bellow is proof, on Latency windows, first two is Balanced Power Plan.

And screenshot when playing a game, System uptime so far is 5H20min.

Yesterday was around 10h, before that was when i did the stability test with new tweaks.

Previously temps would go for 80ºC when doing MemTest and 61+ when playing games, i thought it could be new BIOS, but it wasn't.

Edit: Temps are on Kraken x62 on very silent setting for Fans.
 

Attachments

·
Registered
Joined
·
49 Posts
Question for people who've messed with BCLK, it seems there's a certain bclk (last one I tried was 101.8, knew for a fact 102 didn't work) that once passed over will just completely disable my m.2 sata MX500 drive. My nvme 970 evo and SATA HDD are just fine at that same value (970 is my boot drive and i can get into windows with no problem). My 970 is in the slot that shares pcie lanes with the GPU, the MX500 is in the bottom m.2 pcie/SATA slot. I'm wondering if this is a behavior of my drive or if the C7H disables that port above a certain bclk.
 

·
Realist
Joined
·
376 Posts
Tested my dual rank kit some more last night, tried for 3466 14-14-14-28-42 272 1.41v DRAM 1.0v SoC which booted fine but was getting errors really fast in memtest, more than likely I needed more dram voltage, but 3400 with the exact same settings is about 1100% memtest stable which is cool. I ran a quick and dirty test at 3400 with 1.4v vdimm instead of 1.41 and it threw one error out of 10 tests @150%, so that little bump in voltage seems to have helped. Might try 3466/3500 at 1.43-45v or so next and see if that's possible.
Can you elaborate on what all your BIOS settings are for getting 3400 stable? I get errors with the same settings as far as RTC goes, you mention 1.41V on the memory but your HWiNFO in the screenshot suggests it's at 1.385V?
 

·
Registered
Joined
·
49 Posts
Can you elaborate on what all your BIOS settings are for getting 3400 stable? I get errors with the same settings as far as RTC goes, you mention 1.41V on the memory but your HWiNFO in the screenshot suggests it's at 1.385V?
Honestly I'm not sure why hwinfo is showing that voltage on DRAM, I have both boot voltage and DRAM voltage at 1.41v in bios; soc at 1.0.
Here's all the relevant BIOS settings I don't have on auto (on bios 0804 btw):

(extreme tweakers tab)
DOCP Standard
PE2 (PBO on, 1000 on all values, 10x scalar)
core voltage -100mv offset

(tweakers paradise)
vddp voltage .855

(digi+ power)
DRAM current capability 100%
DRAM power phase control extreme
DRAM vboot voltage 1.41v

every other value is in the attached shot (recommended value used) except I have procODT at 68.6ohm and DRAM voltage at 1.41.
 

Attachments

·
DRAM Obsessor
Joined
·
1,033 Posts
Question for people who've messed with BCLK, it seems there's a certain bclk (last one I tried was 101.8, knew for a fact 102 didn't work) that once passed over will just completely disable my m.2 sata MX500 drive. My nvme 970 evo and SATA HDD are just fine at that same value (970 is my boot drive and i can get into windows with no problem). My 970 is in the slot that shares pcie lanes with the GPU, the MX500 is in the bottom m.2 pcie/SATA slot. I'm wondering if this is a behavior of my drive or if the C7H disables that port above a certain bclk.

That is 100% the behavior of the m.2 Sata SSD. I had the same problem when I bought an 850 EVO m.2 SSD, my 960 pro works amazingly in it even at 106 bclk as well as my 850evo SSD that are connected with a sata3 connection. It just seems like there is a problem with M.2 SATA connections and base clock. It will quickly corrupt any and all data on the drive as well.
 

·
DRAM Obsessor
Joined
·
1,033 Posts
Ahh I figured. Thanks for the confirmation. I guess I'll leave bclk alone until I feel like grabbing a standard SSD.
SSDs are getting stupid cheap now...If you want a really good SSD comparable to Samsung EVO for half the price look for "WD Black" which is western digitals higher performance SSD and has been getting really good reviews as well as speed. Im looking to get a 1tb soon unless the Samsung EVO 1tb drops in price before then. 1TB WD Black cant beat comparable performance for 100$ cheaper than the 970 evo. Im holding out since they will be realeasing the newer technology for SSDs soon and the prices will plummet for the older stuff. If you only need 500gb id go for the 960 pro though. Cheaper than 970 evo and the WD Black 500gb and faster than both with a longer warranty
 

·
Realist
Joined
·
376 Posts
Honestly I'm not sure why hwinfo is showing that voltage on DRAM, I have both boot voltage and DRAM voltage at 1.41v in bios; soc at 1.0.
Here's all the relevant BIOS settings I don't have on auto (on bios 0804 btw):

(extreme tweakers tab)
DOCP Standard
PE2 (PBO on, 1000 on all values, 10x scalar)
core voltage -100mv offset

(tweakers paradise)
vddp voltage .855

(digi+ power)
DRAM current capability 100%
DRAM power phase control extreme
DRAM vboot voltage 1.41v

every other value is in the attached shot (recommended value used) except I have procODT at 68.6ohm and DRAM voltage at 1.41.
Damn, you either have a very good IMC or 0804 is much better than 1101 for this. I'd bet on the chip though, I can't get a -100mv offset without crashing while trying to boot into windows.
 

·
Banned
Joined
·
403 Posts
Nice, also got a nice one. Running a little hotter than Id like but will work. Maybe the 360 would help a little or something....
 

·
DRAM Obsessor
Joined
·
1,033 Posts
Finished my memory tweaking and got my final setup at 3562mhz stable and the performance is beyond what I expected. If I had to say then Id say it runs equivalent to [email protected] at the same timings. Once im done testing a few more things ill have an AMD CBS optimization for taking everything off of Auto and enabling the correct setting which in my case resulted in some bandwidth gains and responsiveness to the system as a whole. Nothing major but any gain is good and the most important part is there is no way the bios can change itself from boot to boot if there are no settings left to chance on Auto.




 

·
Registered
Joined
·
49 Posts
Damn, you either have a very good IMC or 0804 is much better than 1101 for this. I'd bet on the chip though, I can't get a -100mv offset without crashing while trying to boot into windows.
It might be my chip, though I've tried lowering the offset farther and it refuses to even attempt a boot lower than -100mv, but it's been rocksteady there for like 2.5 months through all of my ram testing. I haven't tried any of the newer BIOS yet because I was hoping with the new AGESA there'd be dual rank improvements reported.
 

·
Registered
Joined
·
1,682 Posts
Finished my memory tweaking and got my final setup at 3562mhz stable and the performance is beyond what I expected. If I had to say then Id say it runs equivalent to [email protected] at the same timings. Once im done testing a few more things ill have an AMD CBS optimization for taking everything off of Auto and enabling the correct setting which in my case resulted in some bandwidth gains and responsiveness to the system as a whole. Nothing major but any gain is good and the most important part is there is no way the bios can change itself from boot to boot if there are no settings left to chance on Auto.




Damn nice. And Those low temps 17c on ddr on that voltage
 

·
Meddling user
Joined
·
7,423 Posts
For members with 32GB setups :) .

UEFI 1103, 2700X PE: Default (Some test runs with PBO: Enabled, not PE1-4, etc used), F4-3200C14Q-32GVK.

For a fare length of time I have been using same RAM/CPU between C6H/C7H. I have seen results of upto 3400MHz lasting 9hrs in RAM stability programs, but some what consistency of stability has been thwarting me POST to POST.

Here is some testing run over the course of 2-3 days. The files are ordered by date, most recent setup at top is passing stability testing for a 6th+ warm POST :D .


As you can see from varying % tested it was not the length of testing which was really the issue. An orange box highlights a test run of 12000%+, it was more so to do with on my 6th consecutive warm POST I'd lose stability.

From previous and shown tests ProcODT 48 always came out as best with combined HW, regardless of UEFI used and or which SN# of dimm in which slot. CAD Bus 24 20 20 24 always again seemed solid for me. RTT was odd, several setups seemed sound, none IMO got rid of the POST to POST issue I was seeing. In the end the final setup seems as if is going to use the default RTT settings.

If I changed ProcODT from 48 and or deviated from CAD Bus/RTT settings that aided stability I had issues very quickly, see the red boxed files. At the blue lined point I moved around the dimms within slots. Again really can't say it helped me IMO. The break seems to have come from matching VDDP and CLDO_VDDP. Usually I use default values manually set, VDDP: 0.900V, CLDO_VDDP: 950mV.

Below is the ~13000%.


That test was not on a warm/setup POST, but a POST from having pressed reset, as reboot from OS once ending prior test the board had hung at 7A. If you ref the list of screenies setup 1.043 1.365 0.687 48 24 20 20 24 VDDP 885 CLDO 885 lasted for 5 POSTs, then the test above, one more warm POST and then rig did Q-Code: F9 on next warm POST.

Using VDDP 0.915V & CLDO_VDDP 915mV seems to have sorted this destabilisation after 6th consecutive warm POST.


Rig is now on it's 8th warm POST test run :D . Some of you may have noted the the filenames changed at a point not to have PBOEP in them, those test are without PBO: Enabled. I shall be re-enabling in next set of tests. I think I'll also ref some past tests, to see what lowered SOC/VDIMM/timings I have used for 32GB @ 3333MHz in the past and see if they last for 6+ consecutive warm POST runs with VDDP & CLDO_VDDP tweak.

@crakej

Nice :thumb: .
 

·
Registered
Joined
·
49 Posts
Forget everything I said about 3400 being stable. Whew boy did it start destabilizing hard. I can't even get it to go more than like 20 seconds in memtest without throwing an error and it crashes everything, sometimes programs other times full system. I changed absolutely nothing, seems it was stable for that one post and went to **** quick on subsequent restarts. Computers are a mystery to me. lol back to 3200 I go for now
 

·
Registered
Joined
·
1,456 Posts
Guys, is it possible to have issue with SX8200 NVMe SSD (in M2 bottom slot -second, but directly linked to CPU) and two cards Vega 56? My friend cant install WIndows on PC, always he got error on start of Win10 installation process (attach.)
 

Attachments

4721 - 4740 of 12401 Posts
Top