Clevo Overclocker's Lounge

The guy says the amplifier allows for a longer period of upgradability, but totally glosses over the fact that anyone using a flagship graphics card will run into a CPU bottleneck pretty quickly with a piece of 40W BGA garbage.

A desktop CPU may only be able to be upgraded for 2 cycles, but a 90W desktop CPU will last a lot longer before it becomes a bottleneck for a GPU.

This is all not even considering the fact that replaceable CPU/GPU make repair much easier.

That's my take on it at least. Maybe this guy knows something I don't...


Sent from my iPhone using Tapatalk

 

And what if you want to use your filthy soldered Crap for more than just gaming and web mail and have to use this piece of crap for more than 3 years down the road. Not all replace their computers like they do with their underwear. Be locked down with this BGA filth is a tragedy!!

 

@Mr. Fox I am still having some instability issues with the ram at 3000mhz after a few days of playing around. I was wondering in your bios you had the option to change tREFI and I do not. Is there a different program or bios other than the prema bios that I could use to have those additional options to change tREFI?

 

Did yours come with the Gskill 3000? This is making me concerned as I just ordered one with it, and hoped I would not have any issues.

 

No, I bought from sager and upgraded so do not have access to the prema bios right now but was just trying to figure out a way to change more settings than are offered in the bios.

 

I do believe that others who bought it with there machines were having issues getting it stable but don't quote me on that because I am not positive. Just thought I saw some others on here having issues as well but do not know what bios they were running.

 

aside from the prema bios, theres only xtu or thaiphoon burner to change ram timings.

Sent from my Huawei Mate 8 NXT-AL10

 

which would be the better/easier to work with?

 

easiest to use is definitely xtu. plus, its free. thaiphoon burner goes deep in its settings and costs money. but also provides way more options to change.

im thinking about getting a TB license sometime soon, since im slowly reaching the limits of what i can tweak via xtu/throttlestop/prema bios without having succeeded in getting my ram stable at 3000 mhz. im suspecting i just need a bit more dimm voltage and TB would be the only option to get that definitely cheaper than trying different ram!

Sent from my Huawei Mate 8 NXT-AL10

 

Thanks. I am not sure really what to start tweeking first as I don't have any experience with it. In XMP 1 settings at custom - 1.25 volts - Stock CPU settings for now until I figure out the stability I have run AIDA64 for several hours, made 6 passes of TM5 with no errors, ran memtest86+ for 20 hours total with no errors, Ran several 3D mark and played about 7 hours of BF 1 but my 2 crashes (blue screens) have happened just while I have been working on spreadsheets for work. Could you give me any pointers on what I should concentrate on changing in the settings first?

 

What bsod choose or error are you getting? Are you sure it isn't some other instability (sorry, I play devil's advocate to ensure the stated culprit is the culprit). Is your cache multiplier at stock or overclocked? Please take no offense to the question as an insult to your knowledge or expertise...

Sent from my SM-G900P using Tapatalk

 

To answer your question I do not know what the error was if it happens again I will take a look. Everything is at stock settings except for the ram.

 

Edit.
Fire up Who Crashed or similar and see in the logs.

 

Looks like it was not a ram issue then.

On Fri 12/16/2016 8:36:53 AM your computer crashed
crash dump file: C:\Windows\Minidump\121616-10250-01.dmp
This was probably caused by the following module: syntp.sys (0xFFFFF80720360BF4)
Bugcheck code: 0xD1 (0xFFFF9F8466328008, 0x2, 0x0, 0xFFFFF80720360BF4)
Error: DRIVER_IRQL_NOT_LESS_OR_EQUAL
file path: C:\Windows\system32\drivers\syntp.sys
product: Synaptics Pointing Device Driver
company: Synaptics Incorporated
description: Synaptics Touchpad Win64 Driver
Bug check description: This indicates that a kernel-mode driver attempted to access pageable memory at a process IRQL that was too high.
This appears to be a typical software driver bug and is not likely to be caused by a hardware problem.
A third party driver was identified as the probable root cause of this system error. It is suggested you look for an update for the following driver: syntp.sys (Synaptics Touchpad Win64 Driver, Synaptics Incorporated).
Google query: Synaptics Incorporated DRIVER_IRQL_NOT_LESS_OR_EQUAL

 

This can be the driver, but you are correct that it can be a memory error (now that we know the error). Try to rule the driver out first, then return to the memory. Search without the driver on Google what the common memory cause of this are (I forgot off the top of my head if it is a voltage or timing issue)...

Edit: usually this is a driver issue though...

Sent from my SM-G900P using Tapatalk

 

I get the same BSOD with the memory errors. It points to the synaptics driver, but the goofed up Clevo BIOS memory tables seems to be causing it. Thaiphoon Burner does not show tREFI values in the XMP profiles. Possibly because XMP does not include tREFI. (I don't know this... only know that nothing is there in the XMP profile for it.) The only way I can change that is with @Prema BIOS and Custom Memory Profile. XTU might have it. I don't know because I don't have XTU installed. If it has the tREFI setting that might be the only way you can change it.

 

I mean I have read that this driver has been a problem before on Clevo machines, but that was fixed. But I don't remember the fix.

 

Thanks, are you still getting the same BSOD with memory errors or has it gone away for you with changing the tREFI setting?

 

I see that prema now has our computers listed on his website but has not added the prema bios yet. Maybe I will get lucky in the future and get a chance to download it (fingers crossed).

 

Ya I looked on the website and did not see a upgrade from the current driver I have; however, I downloaded it and uninstalled/reinstalled.

 

No, I do not have any memory related problems that I can identify when using my manual settings. It's very stable now. But, it's not stable at all if I just set it to use XMP and go.

The exception is the Kingston 4GB sticks. Those do work right set to XMP. I don't have a clue why, but I am guessing it has something to do with them only being 4GB capacity.

It is interesting that the Eurocom Tornado F5 boots and runs XMP-3200 G.SKILL 4*16GB with no problems. Maybe @Prema can examine what MSI did to the memory part of the BIOS and do the same to straighten out the Clevo memory mess.

 

Ok well I guess I just still need to do a little more tweeking. I did have it set in custom when it crashed so I will go down to XMP2 settings at custom and see if I run into any issues. Will also download XTU and play around. Thanks all for the help!!

 

Using 1.350V for memory seems to help increase speeds, including cache speed, and lowers latency. Using the DDR4 "standard" of 1.200V hasn't ever been ideal as far as I am concerned. That's not enough voltage at high clock speeds. Running CPU and GPU at higher frequencies takes more voltage, so it kind of makes sense on the memory as well. This benchmark is with 1.350V set in the BIOS for my custom memory profile.

The modern obsession with using less voltage doesn't always work as well as the tree-huggers would like us to believe. Hitting some kind of silly arbitrary value should never be the goal. The goal should be to find and use the correct amount, whether that right amount is more or less voltage is totally irrelevant.

 

To build on this, just don't be reckless and memory can take a bit. You should find the data sheet on the chips that the sodimm uses to figure out temp limits for the chips. Then it is figuring out how hot they are running in your case. 1.35V is usually considered safe for ddr4 for what I've seen on xmp profile, etc. Desktops have better air movement which allows for 1.4V (I use more but I have active air flow directly on the chips). Just be safe...

Sent from my SM-G900P using Tapatalk

 

Unfortunately, until Clevo resolves whatever conflicts they have on SMBus, there is no way to know the memory chip temperature. There is no legitimate reason for SMBus Support to need to be disabled in HWiNFO64. Having it enabled caused me no issues on my previous machines (including P870DM-G) and that works fine on the P750ZM as well. You must have SMBus Support enabled to get the DIMM temperature readings. For the purpose of demonstration I re-enabled SMBus Support to get the screen shot below for this post. I launched TM5 and within 1 second my machine froze, then BSOD with MACHINE_CHECK_EXCEPTION bugcheck. The machine would not POST again until it had gone through 4 or 5 power cycles like it is trying to find RAM settings it needs to POST. If I disable SMBus Support everything is fine. Well, not really fine because Clevo needs to fix this problem, but fine in the sense that my machine regains stability.

Anyhow, with SMBus Support disabled, you won't have the memory thermal data shown in the screen shot below.



Edit: And, as expected... TM5 passes with SMBus Support disabled. Very predictable behavior induced by a problem that Clevo needs to fix.

 

Excluding the Ram voltage what is the safe Voltage for 6700k on these machines.

 

High temp is your enemy before voltage will be a problem

 

Point taken. I was meaning generally, but this is an excellent point that the current situation prevents accurate tuning/optimizing...

Sent from my SM-G900P using Tapatalk

 

id recommend lower than 1.4V for 24/7 usage and max. 1.45V for short benches.

Sent from my Huawei Mate 8 NXT-AL10

 

If you mean CPU voltage, I do not believe these systems are designed in such a way to allow enough CPU voltage to cause damage. @Johnksss@iBUYPOWER has mentioned this a number of times. That's why we cannot overclock quite as high as the same CPU would if it were installed in a proper desktop enthusiast motherboard. Clevo capped the voltage limits too low (plus there is no VCCIN adjustment) for us to fully stability the CPU beyond 4.9GHz. We would need the limit raised to more like 1.550V to 1.600V core and 2.200V max input voltage to go higher. So, the good news is the only thing you have to be concerned with is tuning to avoid Current Limit Throttling and voltage-induced thermal throttling.

 

And you have a decent drop in voltage when processor is under load so your +1.4v will be a lot lower under heavy use Temp will stop you long before voltage will be a problem.

 

Good deal I just noticed online some saying to keep it below 1.4 as Jaybee mentioned and others saying to stay under 1.35 so was just curious.

 

And, as expected... TM5 passes with SMBus Support disabled. Very predictable behavior induced by a problem that Clevo needs to fix.

 

Info for you Read carefully what is written in the green box Same color as the Smilies

http://overclocking.guide/the-risks-of-overclocking/



"Intel's recommended voltages for the Skylake CPU from their 6th Generation Datasheet http://www.intel.com/content/www/us...ktop-6th-gen-core-family-datasheet-vol-1.html You are welcome to follow Intel's specifications, but for VCore, I don't recommend running over 1.5v even though Intel supports up to 1.52v"
http://www.tweaktown.com/guides/748...-intel-skylake-overclocking-guide/index5.html

 

Agreed... the beauty of not being shackled by BGA filth is you don't have to care all that much. If you draw a short straw in the silicon lottery... you replace it. If you wear it out from too much overclocking (hasn't happened to me so far)... you replace it. If you kill it on accident because you did something stupid... you replace it. If something better is available (i.e. Xeon or X vs K, or XM vs QM, perhaps Kaby Lake vs Sky Lake) and you want one... you replace it. And, since most K-series desktop CPUs are very cheap (new or used) in the grand scheme of things, it is fairly painless compared to the price of a BGA motherboard with the better CPU and better GPU options.

 

Thanks, I get it been overclocking other processors on water and with noctua nhd15 for a while but was always able to get decent overclocks with fairly low voltage. 1.5 sounds pretty high to me which in our case does not matter without a better cooling solution. I have just always tried to stick below 1.35 but it is good to know they will withstand that much voltage.

 

On the other hand I don't have experience with overclocking ram and without a decent bios it makes it difficult to fine tune the CPU voltage for a good overclock.

 

You'd be surprised to know that most of @Johnksss@iBUYPOWER 's earlier high scores were with the stock BIOS.

 

Neither more nor less. Just the way it always should be. BGA is born to die!! Soldered Trashware should only be used in tablet, phones and equipment under 14 inches. Not in normal laptops/desktop rigs!!

 

Interesting I just see the voltage getting to high for my likes with not being able to really lock it in at a voltage that stays consistent.

 

Coincidence aka almost same time, driver or memory problem here too?
http://forum.notebookreview.com/thr...s-owners-lounge.795090/page-467#post-10412876

 

Indeed. The value in the unlocked BIOS is about 30% CPU overclocking and the other 70% has to do with restoring critical features for basic but full functionality, and memory configuration options, that are hidden from customers in the stock BIOS. However, anyone that is using the EC that came with their machine can expect some really severe (and super-disappointing) current limit throttling problems unless they have the same older EC that @Johnksss@iBUYPOWER and @Meaker@Sager had. I'd never even consider buying one of these machines if I had to use it forever with the stock BIOS because functionality is too crippled for words and the current limit throttling alone is grounds for an RMA in my mind. Paying big bucks for something castrated by firmware cancer that can't hold its CPU clock speeds 100% under load is totally unacceptable. It is truly sinful that Clevo released the newer firmware versions to the public considering how severely screwed up they are.

It certainly might be. I replied to the person at the other end of the link. Synaptic drivers have always been goofy and called out in stop errors with an unstable overclock. Even back in the 2920XM days that was happening, but the unstable XMP/memory overclocking thing makes it worse than I ever remember from the past. Considering the type of machine (BGA) it would not surprise me if an overly-aggressive CPU under-volt is causing the instability.

 

I already put out a newer version.

 

Hmmm? What? New EC version? Where?

Is it going to fix all of the current limit throttling for non-Prema eligible owners like they deserve for it to be fixed?

 

So just quickly playing around with XTU tonight I set static voltage to 1.2 ran 4.5 X 4 Cores and 4.2 cache. Playing BF1 for 30 minutes at 4.5/4.2 and +100 GPU got the best average FPS I have gotten so far which was nice. It seems at static 1.2 voltage the voltage shows to be 1.206 to 1.409 assuming CPUID is accurate. Not very static at all!!

 

Nice. Good job, bro.

Sky Lake voltage has to be measured under load, not idle. It's never truly static even with a static voltage setting, except for when it's under full load. But high idle voltage does not contribute much to thermal issues at all, and it helps keep the system stable when clock speeds are erratically moving up and down based on load/demand.

 

Ok, I do not remember seeing this with my haswell, was static when I set it.

 

Makes sense though it is more difficult to find that exact voltage when it is static.

 

Yes, it takes some getting used to. Adaptive voltage was not very good for overclocking with Haswell, but the longer I use Skylake the more effective adaptive voltage seems to be. I had to re-learn a few things because what I knew from the past no longer applied 100%.

 

Run bench and find your <hidden >load voltage. You're welcome Run as much as you like wPRIME v1.5 - CINEBENCH_R11.5 - CINEBENCH_R15.03 - 3DMark11

"During normal CPU operation, the CPU load will either go up or go down, and the auto CPU voltage goes in the opposite direction of the load. When the load is higher, current is also much higher, and staying at the same VCore would result in much higher power consumption since power is heavily dependent on voltage and current. To maintain the same power window and compensate for low quality VRMs, the VCore is dropped when current/load increases"