2700x Performance in PBO (Precision Boost Overdrive) only - Page 6 - Overclock.net - An Overclocking Community

Forum Jump: 

2700x Performance in PBO (Precision Boost Overdrive) only

Reply
 
Thread Tools
post #51 of 77 (permalink) Old 01-14-2019, 04:53 AM
Meddling user
 
gupsterg's Avatar
 
Join Date: Jan 2015
Location: Lurking over a keyboard
Posts: 7,160
Rep: 774 (Unique: 354)
Quote: Originally Posted by constructorx View Post
I think so yes, I will check. I believe that is what the RMC instructs to do. What is the effect/impact of this?

I have just read:

"The effects of the Gear Down Mode option appear to be a little more complicated. While it is enabled, setting the "command rate" option for memory timings is pointless; it will always effectively be "2T" (although the company notes it's not quite the same thing—1T versus 2T command rates may still be worth exploring on their own). Disabling GDM will allow memory overclockers to manually set the DRAM command rate to "1T" and realize the full benefits of doing so. Tweakers have known for a long time that a 1T command rate can have huge effects on memory performance, and the Ryzen platform is no exception. However, if your memory can't run a 1T command rate at its spec'd clock rate, or if you're overclocking and can't hit a high clock with a 1T command rate setting, you could just be making your system less stable by disabling GDM."

My memory kit is 1T by default so maybe no benefit?
Dunno if that info from Techreport is correct.

I can't say it's "huge effects".

GDM allows the RAM to use every other clock edge for CMD/AMD, so basically like 2T, link.

Spoiler!


Due to this I reckon odd CAS is rounded to even when GDM is on.

This older article had what I thought was great nugget on 2T.

Quote:
The use of 2T timing is not entirely without its drawbacks. The clock enable (CKE), on-die termination (ODT) and chip select (CS) signals cannot use 2T timing, as it is critical which clock edge samples these signals.
Currently for me 2T and or GDM did not help gaining >3466MHz, only a CPU with differing IMC on same RAM/MOBO did.
gupsterg is offline  
Sponsored Links
Advertisement
 
post #52 of 77 (permalink) Old 01-14-2019, 06:06 AM
New to Overclock.net
 
umeng2002's Avatar
 
Join Date: Jul 2010
Location: Florida
Posts: 3,133
Rep: 170 (Unique: 106)
Update, it's not the GearDown mode. Computer just froze while doing lightweight CAD work.

I'm going to start to loosen the sub timings.

CPU
AMD Ryzen 2700X
Motherboard
Asus Prime X470-Pro
GPU
EVGA GeForce RTX 2070 XC Ultra
RAM
TeamGroup T-Force 16 GB (2x8) Pro Dark (B-die TDPGD416G3200HC14ADC01)
Hard Drive
Samsung 840 EVO 250 GB
Power Supply
Seasonic Focus Plus Platinum SSR-750PX
Cooling
Corsair H80i (not V2 or GT)
Monitor
LG 34UC80-B
Keyboard
Logitech G413
Mouse
Logitech G503 RGB
Audio
Creative SoundBlaster Z (OEM)
▲ hide details ▲
umeng2002 is offline  
post #53 of 77 (permalink) Old 01-14-2019, 01:57 PM - Thread Starter
New to Overclock.net
 
Join Date: Jan 2019
Posts: 81
Rep: 0
Quote: Originally Posted by umeng2002 View Post
The fast preset tells you to disable GearDown mode, but the DDR4 spec turns it on past 2666 MHz.
I think it is different for every specification of kit. My kit (F4-3200C14D-32GTZ) is a dual rank 32GB kit consisting of 2 sticks of 16GB.

The RMC 'fast' settings for my kit state:

  • Power Down: Disabled
  • Gear Down: Enabled
  • BGS: Enabled
  • Command Rate: 1T

I have tried (albeit quickly) to test if I could see any difference and although I got a few points higher in Cinebench, I am not 100% it is because of the gear down disabled. I would need to fully understand what is happening and test again. With the settings entered exactly as RMC states I am quite quick and not complaining.

Last edited by constructorx; 01-14-2019 at 02:03 PM.
constructorx is offline  
Sponsored Links
Advertisement
 
post #54 of 77 (permalink) Old 01-14-2019, 02:51 PM
New to Overclock.net
 
umeng2002's Avatar
 
Join Date: Jul 2010
Location: Florida
Posts: 3,133
Rep: 170 (Unique: 106)
Well, setting dual rank in the Calculator will change the recommended options.

CPU
AMD Ryzen 2700X
Motherboard
Asus Prime X470-Pro
GPU
EVGA GeForce RTX 2070 XC Ultra
RAM
TeamGroup T-Force 16 GB (2x8) Pro Dark (B-die TDPGD416G3200HC14ADC01)
Hard Drive
Samsung 840 EVO 250 GB
Power Supply
Seasonic Focus Plus Platinum SSR-750PX
Cooling
Corsair H80i (not V2 or GT)
Monitor
LG 34UC80-B
Keyboard
Logitech G413
Mouse
Logitech G503 RGB
Audio
Creative SoundBlaster Z (OEM)
▲ hide details ▲
umeng2002 is offline  
post #55 of 77 (permalink) Old 01-15-2019, 02:27 PM - Thread Starter
New to Overclock.net
 
Join Date: Jan 2019
Posts: 81
Rep: 0
Quote: Originally Posted by umeng2002 View Post
Well, setting dual rank in the Calculator will change the recommended options.
My memory kit is dual rank, so I have to enter that. I have no choice.
constructorx is offline  
post #56 of 77 (permalink) Old 02-05-2019, 04:31 AM
New to Overclock.net
 
Join Date: Jan 2019
Posts: 49
Rep: 0
All,

I am really curious about something here.

In following along here, I am able to get decent Cinebench scores on the first run (1930+) when setting my vCore to a static override (1.38+). However, on the 2nd run (or more), my 2700x system locks up.

(Note that I do not waste any time in between executions... I run the 2nd test as soon as the button is enabled and will allow)

Is anyone else able to get consistent consecutive Cinebench executions when using PBO + vCore override? (To me, if you can execute 5-6 times without locking that means things work ) Is this a known issue and/or something to ignore?

It's really driving me nuts. I don't know if maybe I have something misconfigured?

FWIW, I am able to run Cinebench over and over when vCore is set to AUTO or using the offset. However, my scores are in the 1800s when doing this.

Thanks in advance for any assistance/insight you can lend.

EDIT:

Adding a little more context. It seems that the root of this problem is that the heat of the Ryzen chip continues to increase with each and every Cinebench execution. That is, there is no "limit" and it continues to climb until the controls kick in to shut off the machine (even though the machine doesn't shut off... it just freezes). That is my hunch, at least.

MORE EDIT:

The best I can do currently is run two executions of Cinebench consecutively. The third one locks up. If I wait a little bit after the second execution (I wait until the temps get into the 30s), then I am able to execute a 3rd time.

Memory: HX433C16PB3K4/64 @ 3000 14-16-14-17-28 1.3v / CPU: 2700x Ryzen @ 4.275 1.375v & 1.025 vSoC / Motherboard: msi x470 Gaming Pro / Power Supply: 650w Snow Silent / HDs: 250gb Samsung 970, 500gb Samsung 970, 4x 500gb 850 Samsung EVO (1tb mirror RAID).

Last edited by Mike-EEE; 02-05-2019 at 04:52 AM.
Mike-EEE is offline  
post #57 of 77 (permalink) Old 02-05-2019, 05:01 AM - Thread Starter
New to Overclock.net
 
Join Date: Jan 2019
Posts: 81
Rep: 0
Quote: Originally Posted by Mike-EEE View Post
All,

I am really curious about something here.

In following along here, I am able to get decent Cinebench scores on the first run (1930+) when setting my vCore to a static override (1.38+). However, on the 2nd run (or more), my 2700x system locks up.

(Note that I do not waste any time in between executions... I run the 2nd test as soon as the button is enabled and will allow)

Is anyone else able to get consistent consecutive Cinebench executions when using PBO + vCore override? (To me, if you can execute 5-6 times without locking that means things work ) Is this a known issue and/or something to ignore?

It's really driving me nuts. I don't know if maybe I have something misconfigured?

FWIW, I am able to run Cinebench over and over when vCore is set to AUTO or using the offset. However, my scores are in the 1800s when doing this.

Thanks in advance for any assistance/insight you can lend.

EDIT:

Adding a little more context. It seems that the root of this problem is that the heat of the Ryzen chip continues to increase with each and every Cinebench execution. That is, there is no "limit" and it continues to climb until the controls kick in to shut off the machine (even though the machine doesn't shut off... it just freezes). That is my hunch, at least.

MORE EDIT:

The best I can do currently is run two executions of Cinebench consecutively. The third one locks up. If I wait a little bit after the second execution (I wait until the temps get into the 30s), then I am able to execute a 3rd time.
My 24/7 settings give me 1972 Cinebench (between 1965-1972) when warmed up and have been solid for weeks. I tested my setting with both memtest and Prime95.

What are your temps and voltages under load in HWInfo?
constructorx is offline  
post #58 of 77 (permalink) Old 02-05-2019, 05:38 AM
New to Overclock.net
 
Join Date: Jan 2019
Posts: 49
Rep: 0
Quote: Originally Posted by constructorx View Post
My 24/7 settings give me 1972 Cinebench (between 1965-1972) when warmed up and have been solid for weeks. I tested my setting with both memtest and Prime95.

What are your temps and voltages under load in HWInfo?
They have literally been all over the place, LOL. So much so that I am starting to think that the calculations that I have been getting from Ryzen Calculator have been fundamentally wrong (e.g. CJR vs MFR -- even though there's a CJR in my ID when I look at it in Typhoon).

To answer seriously, though: the best results I have gotten are by using 1.4 vCore @ LLC 2. This allows me to run Cinebench twice. The third time results in a freeze unless I give it some time to cool down.

Here are the sample temperatures from an execution set:
1) 74, score 1940
2) 75, score 1940
3) 77 -- freeze

As for memory, I have used anywhere between 1.28 to 1.45. The same goes for SoC. I have been able to set to .9 all the way to AUTO (which is 1.15 and from what I understand WAY too much). Nothing seems to keep the temperatures from climbing, resulting in the frozen state.

The only way I can get legit testing is to use offset VCore or by setting it to AUTO (or something greater than 1.4 @ LLC2).

A consideration: I am using 4 sticks instead of 2.

EDIT: I forgot to provide vCore under load. The values that I am seeing are 1.406 and 1.412. It would seem that HwInfo keeps me from getting the 2nd full run as when I attempt to run two consecutive executions, it freezes on the 2nd with HwInfo open (rather than on the 3rd without it opened).

Memory: HX433C16PB3K4/64 @ 3000 14-16-14-17-28 1.3v / CPU: 2700x Ryzen @ 4.275 1.375v & 1.025 vSoC / Motherboard: msi x470 Gaming Pro / Power Supply: 650w Snow Silent / HDs: 250gb Samsung 970, 500gb Samsung 970, 4x 500gb 850 Samsung EVO (1tb mirror RAID).

Last edited by Mike-EEE; 02-05-2019 at 05:51 AM.
Mike-EEE is offline  
post #59 of 77 (permalink) Old 02-05-2019, 09:10 AM - Thread Starter
New to Overclock.net
 
Join Date: Jan 2019
Posts: 81
Rep: 0
Quote: Originally Posted by Mike-EEE View Post
A consideration: I am using 4 sticks instead of 2.

EDIT: I forgot to provide vCore under load. The values that I am seeing are 1.406 and 1.412. It would seem that HwInfo keeps me from getting the 2nd full run as when I attempt to run two consecutive executions, it freezes on the 2nd with HwInfo open (rather than on the 3rd without it opened).
That is a lot more voltage than I am pushing. I think the best approach is to have a separate profile on your motherboard (of it supports saving profiles with different settings) and start with everything set to default. Then get your memory timings as good as you can. Then enable PBO with a slight undervolt. This is how I get the best performance. I have my CPU Core Voltage set to 1.3V which under load pushes at max between 1.325 to 1.35 (depending on my settings). I take your point about 4 sticks of memory, that is supposed to be not as effective when overclocking.
constructorx is offline  
post #60 of 77 (permalink) Old 02-05-2019, 10:07 AM
New to Overclock.net
 
Join Date: Jan 2019
Posts: 49
Rep: 0
Quote: Originally Posted by constructorx View Post
That is a lot more voltage than I am pushing. I think the best approach is to have a separate profile on your motherboard (of it supports saving profiles with different settings) and start with everything set to default. Then get your memory timings as good as you can. Then enable PBO with a slight undervolt. This is how I get the best performance. I have my CPU Core Voltage set to 1.3V which under load pushes at max between 1.325 to 1.35 (depending on my settings). I take your point about 4 sticks of memory, that is supposed to be not as effective when overclocking.
Yeah, starting over from defaults is a good idea. Thank you for your feedback, @constructorx . It is very much appreciated. YOU ARE THE OASIS IN MY DESERT OF GRIEF!!!

Part of the problem has been the motherboard, which is actually great except for the fact that I have spent six full weeks trying to ensure I get my money's worth. Of course, having spend six weeks on anything you sort of lose that argument, doh.

I have indeed tried 1.35 and lower. At that level, Windows will not boot, and if it does, a blue screen follows closely after. 1.36 is the best I can do, and at that point, Cinebench no doubt will freeze and HCI MemTest reports memory errors. The same for 1.375 but longer Cinebench executions and less memory errors (but still... memory errors).

It's gotten so bad that I thought it was the memory, even though I checked it with MemTest when I first got it and it turned out OK at 2400, its default speed.

I went ahead and tried 2900 a few days ago and it was OK there, using the XMP Profile 1 setting.

From there, I can bump it up to 3000 and 3066 no problem. This was the same experience I had with another chipset I was looking at (you can see both of them in my sig -- take advantage of my laziness LOL... I ended up with the Hyper-X set).

However, going beyond that is the source of all my ails. I have a 3300 rated chipset but the best I can do is 3133, and even then it's memory errors after testing for a few hours -- insanity!

And so yeah... that four-chip set is what is taking me for a massive ride of failure here. I have been tempted to simply remove 2 chips out of my motherboard to see how fast I can get to 3200 w/o errors, but I am sure I can do it. In fact, I was able to get 3200 to POST no problem w/ only 2 chips in. With 4 in, I cannot get it to post unless I put the vDRAM to 1.45 (maxed out, essentially).

Memory: HX433C16PB3K4/64 @ 3000 14-16-14-17-28 1.3v / CPU: 2700x Ryzen @ 4.275 1.375v & 1.025 vSoC / Motherboard: msi x470 Gaming Pro / Power Supply: 650w Snow Silent / HDs: 250gb Samsung 970, 500gb Samsung 970, 4x 500gb 850 Samsung EVO (1tb mirror RAID).
Mike-EEE is offline  
Reply

Tags
2700x , Overclock , ryzen

Quick Reply
Message:
Options

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in



Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off