Clevo Overclocker's Lounge

Sure.

 

What possessed you to do such a thing?

 

laziness to configure everything myself and the lack of knowledge as I stil didn't read your optimization thread

 

I meant no offense and apologize if any was given. I didn't realize your specific clocks were set manually. I was simply wondering if the extra voltage was necessary to achieve those speeds and timings or if it would need manual voltage control or, like the ZM series, having to flash a new xmp profile while hunting for stable. In fact, I wasn't referencing your overclock specifically, but was wondering about the ram type overclock generally (meaning how much over stock speeds the ram can go by type, like Samsung B-die versus hynix or the package). I do apologize if it sounded otherwise. Also, I'm running on fumes after a 12 hour day, so if I'm not the most articulate, I apologize for that as well.

To number 3 on your list, have you or anyone else tried memtweakit for the z170 chipset, just to see if it works? Just a thought I had...

4. Could be just variance in the ram kit or batch, which is kind of what I'm trying to figure out (seeing what the range is that the kits give on need for voltage and timings).

5. Once again, I apologize if I wasn't clear in my first post. I was meaning generally, not your specific settings. But it would be nice to get control over the voltage without flashing an XMP (I'd like to be lazy instead of hunting and flashing art the higher voltage). But definitely hear ya.

6. That is quite promising. Still having bought mine yet, probably will have to go with a single 1080 or a dual 1070 because of costs, but it will work...

That sucks to hear that setting is untouchable at the moment (leaving it open and praying for it to someday be open).

I'm on my phone. I'll check out that image from your P570WM here in just a little bit.

And once again, sorry if I gave the wrong impression or offense.

Sent from my SM-G900P using Tapatalk

 

DDR4 RAM prices are going up ;_;

Samsung are the same company who makes all RAM... And they were the ones with Note 7 boom part....Now this is sad...

I don't even have it logged in, manual optimisation.

Is there a thread with optimisation for nvidia drivers? I want to see! I've found a perfect setup lately but am always looking to improve.

Also, geforce doesn't optimise madVR nor other applications, only games. Some pro apps need manual tweaking to give their best.

 

I've made no such thread; I don't know what he's referring to.

If you want, however, there is a website article breaking down most everything about nVidia's drivers. Here: http://www.tweakguides.com/NVFORCE_1.html

 

There was a thread you made that detailed every nVidia Coneroli Panel option if I remember correctly that you even had it in your signature

 

I think you're thinking of this thread Octiceps made perhaps..

http://forum.notebookreview.com/threads/nvidia-geforce-tweak-guide.774045/

 

Oh snap, my bad. Thanks for the clarification bro

 

That gives me a little bit of info, but what I can tell from this alone is that you would kill my ram on SuperPi 32M on latency effects alone. I showed the two kits I had and compared them on MaxxMem2, AIDA64, Winsat, and SuperPi32M not too long ago (geekbench is another that benefits with better ram timings, as does the physics on Time Spy). There is a different memory controller (Mr. Fox couldn't use the same stable settings from his P570WM on his P750ZM and when I threw the 4x8GB kit G.Skill 2133 DDR3 in my P170HM, they could use a lower stable tRFC then my P770ZM could do), etc. Overall, looks like your sticks would do well or, most likely, better than my current sticks. The tRFC effects can vary (but lower is better with it). When I did the G.Skill versus my current ram, I had CL12 13-13-31-278 2T going against my current ram you cited. In AIDA64, the read, write, and copy ALL edged out my Kingston, but lost on latency. It also won in Winsat mem. Where it really became obvious the effects of latency were the lower scores in MaxxMem2 and a two second and change slower SuperPi32M score. The tRAS between kits can vary considerably and is a window, so that doesn't give much. The tRFC there is nice, but I also showed a 278 edging out a 228 in AIDA64, so it is hard to tell without an apples to apples comparison with the IMC and voltages held constant (but lower is better). I don't have the tREFI and tREFIx9 on that screen (both of which higher is better, generally), but as I said, that lower latency is where the real world effect is coming in and where that kit is going to most likely kick my kits ass. Those timings look nice, but as I said, the other variables, from memory controllers, etc., make it a bit harder to compare.

Also, absolutely correct in saying that tighter timings do not always give the best score. That is where I often use SuperPi32M to see which timings (or combo of timings) provide the best score. If you see a trend in the wrong direction... But you know all of this and I apologize if this comes as condescending, I don't mean it to, but wanted to talk it out for others in these pages that don't know as much that may gain a benefit from reading a discussion of this type. Whenever you see me go on like I just did, please ignore it or do not take it as me talking down in any way. It's always for everyone in the community and to help build a knowledge base for someone that is searching for an answer and that may stumble on this (or other) threads. You've been doing this longer than me and at a completely different level than me, so I often defer to you and do welcome corrections.

Meanwhile, for anyone that would like to try MemTweakIt (even though it is an ASUS tool) on their Z170 chipsets just to see if it will work, here is a link to a page that will allow you to download it: http://overclocking.guide/downloads/

Once again, if I previously gave offense, I'm sorry. I did not mean to...

 

It's all good....

 

After using p750dm, p870dm and the new p775dm3, I am realizing that there is a mobo limit in non-p870 mobos.

When cpu and gpu are loaded hardly, cpu clock fluctuate to non-OC level which is 4.0 in p750dm and p775dm3. This issue could not be fixed with prema in p750dm. I am not sure prema bios could fix this for p775dm3 since I dont have prema bios for it.

When cpu was oced P870dm maintained the oced cpu clock whenever but not in p750dm and p775dm3.
I tested alot and I even tested with different and with same cpu.

Not only me but many others experience same in this issue.
I dont want to criticize or blame clevo but since skylake hq cpu can handle 4.0 easily there is not many reason to buy clevo new model for cpu performance. If you want max cpu performance, go for p870DMs.


Ps. I can get good score in 3dmarks with oced 6700k and this is because physics test is giving load to only cpu.
But in real game where cpu and gpu both are working hard, cpu dont keep its clock all the way. It fluctuates alot. There is not many reason to oc cpu for real game since benefit is low and loss is much i.e higher voltage, higher temps.

 

Can you make a video of this?

 

Does anyone know if the prema bios for the P670RP6 has an option somewhere for ASPM control? Or is that feature missing? If someone has the stock bios to a similar model with the 1060, 1070, or 1080, do you see an ASPM setting?

 

Prema fixed it for my P771DM-G, maybe some of you guys remember when I tried and tried and couldn't get it to work, he let me try a new Bios and boom i had steady 4.3Ghz while playing, using Max Performance under energy options in Windows was necessary tho, but still to this day runs like Schmitz katze.

Gesendet von meinem Nexus 6P mit Tapatalk

 

This goes like Schmitzkatze: What is behind it?
https://translate.google.com/transl...atze-was-steckt-dahinter-id35235/&prev=search

"It's like Schmitz's cat!" - If we use this speech, something goes particularly fast or even very simply. But why do we compare such a rapid process with a purring four-legged animal who is known to sleep two-thirds of his life?

That's a new one on me

 

@iunlock - You seem to have played a lot with OC, what are your CC settings?

 

I normally run 46x daily with a -55mv, but now been playing with 47x and getting it stable, which I think it is now since I can game, render, encode and bench at 47x.

 

That is good. I sold my p750dm so I couldnt test the new prema for it. It is very good to hear he fixed it.

 

It's like 4.5 4.5 4.0 4.0 4.3 4.3 ..... so on while graphic test and combined test. Prema told me that is is the mobo low limit and he could fix it but it was when I already sold p750dm.
When I was using p870dm, the max/min clock of hwmonitor or hwinfo was always the same which is the the OCed clokc. But with p750dm and p775dm3, min clock is always 4.0.

 

Any idea why the min clock is 4.0? I also have that problem.
Cpu bench works fine but when cpu and gpu are both under load, cpu clock goes down to 4.0 or somewhere between 4.0~4.6.

The old zm model always kept the oced clock. But dm except for p870, the clock fluctuates.

 

I might need an extended view on what settings you use, the image doesn't work and I can't hold 4.0 clocks on CPU

What is your cache ratio? What are your noise levels (your fans run really fast all the time?)\

Also I want to try to do at least 4.2 stable to 4.4 OC per 2 cores, but so far I must use 2.8 on all cores to be fully stable and it's painful... maybe I really have a setting messed up or something like that

 

A common translation would be: "it runs like a champ now!" or "like a Schnitzel!"

 

OT: originally referring from the "smith's cat" which freaked out and playing "the Flash" when the smith hammered on his anvil and the cat was accidentally nearby.

 

Ah, okay.

@bloodhawk

I got Thunderbolt networking to work. I was using the 20 Gbps cable
It was running around 10 Gbps or 1.13 Gigabyte or 1100 Megabyte max. Low end was about 930+ on average with a few dips into the 850 Megabyte range.
It moved 67.5 gigs in 1 minute 18 seconds.
20 gigs in about 18 seconds. 3 iso files. Stayed in the 1.0+ range

 

Anyone knows why is this happening?

The minimum possible CPU state is set to much higher...

 

Damn that's fast!
Howd you get it to work?
I tried it with the Razer Core but it was super flaky.

 

To perfectly honest. I was running my mouth without using the proper tools. . Thanks to you guys!

After getting the right cable and software, it started working, but I did the easy route and disconnected the Ethernet ports as well. So no internet. And it started working. I have to say, i'm loving this speed!

Darn, was hoping you would get that to work. It was giving me some other ideas for file transfers and clones of nvme drives.

Edit:
Burst speed to Mec drives is insane. 1 GB for about 5 seconds.

Edit:
Internet backup and running and still connected using TB3/Type C

 

I have my 950 Pro running over TB3 right now and its pretty much running exactly as it would plugged in to the m.2 slot (since im using the M.2 Slot for the 1080). I think CrystralDisk mark score was maybe 2-3% lower, but that might have been something running in the background.

Spoiler: PCIe card

Im guessing the connection is TB3 to TB3?

 

Is that card out of the razor core?
And that pcie card running the 950. Is that dual nvme or one side is nvme and the other is sata? Or does the nvme drive get flipped over? And where did you get that pcie card??
Who cares about a 3 percent loss. . That is spectacular in my book!

 

Nope that's the TB3 HDK, basically an open ended version of the Akitio TB3 PCIe box.
I tried the Ethernet port on the Razer Core and it was a pain to get it running, because of razer pathetic drivers.
Even as a eGPU using the Razer Core on non razer systems is a pain. Might as well just mod an Akitio Box or use the HDK in some case.
Razer Cores performance is actually lower in certain cases. Specially then the extra ports are used with the GPU running.

That's a single nvme card with Sata on the other side. Got it off of amazon : https://www.amazon.com/gp/product/B01E8A6SLC/ref=oh_aui_detailpage_o04_s00?ie=UTF8&psc=1
One draw back about this card is that, in order to use the sata side, a sata cable needs to be connected going to the Sata port.

Have this one coming in : https://www.amazon.com/gp/product/B018AOZ9QM/ref=ox_sc_sfl_title_1?ie=UTF8&smid=ATVPDKIKX0DER&th=1 , going to slap in 2 x 1 TB x400's. Not sure if NVME drivers will work with this, as it says mSATA.

Im looking for a PCIe card that has 2 x M.2 NGFF PCIe slots.

 

Guess that knocks the razor off my list of things to test...

Looks like they only have an enclosure or two that will run a nvme drive. And I think only one that will run two.

That one is not nvme, but of course i'm rooting for ya!

And so am I!

 

That's an interesting product. Let us know how it works, I may be interested.

 

It's certainly a nice capacity booster for those who must have more space.

 

This is interesting, I would like to check it personally sometime and see if I can get it to stick.

 

Well, maybe sometime I'll be able to check one of these myself =D.

 

@Papusan @Mr. Fox @bloodhawk



Why does my CPU clock down to 797 MHZ randomly when playing games ?
I tried throttle stop and it was worst

Is it due to the fact that i don't have Prema Bios so i don't have access to power limit 3 & 4 as right now i have maxed out the current limit, VR current limit, power limit 1 & 2
mind you, i have done these adjustments using CCC and have overclocked the CPU on all cores and i have only 1 power adapter.

 

Could be because of the CSTATES or could be the Minimum Processor state in power settings not set to 100%.

Either ways, if it sticks to the turbo frequency under load, then you should be good.

 

Minum power states is 100%

check my edit in the above post.

 

Could be because of the stock BIOS / EC.
I really would suggest against overclocking anything using the CCC or XTU. Would get rid of both, do a full CMOS/NVRAM reset and then install CCC without XTU.
Then use TS to over clock.
TS isn't that easy to understand, but after a bit of effort its one of the best out there.

 

see, with Wprime only no throttling happening, its only when the graphics card is involved



I think its the EC firmware ?

 

Ah if it only happens when the GPU is in use as well, then definitely the EC. Unless the power limits aren't maxed out.
I think @godfafa_kr had a similar problem..

 

Listen to @bloodhawk and do the steps ↑↑↑ . Rater try with TS again. Or this is maybe a EC / bios problem.

 

XTU causes major gremlins in these machines. I'm steering clear of CC and XTU altogether this time round.

 

I'd never seen this benchmark tool before. Interesting. I just ran it on El Cazador with the 4790K @ 4.3GHz and the 980M stock while watching Arrow on Netflix with Mrs. Fox.

It tests a bunch of stuff, including drive speed, and ranks according to percentile against other submissions. Interesting tool for comparing SSD performance as well.

Brother @Johnksss@iBUYPOWER have you ever come across this one before?

Benchmark Download: http://www.userbenchmark.com/Software
Result: http://www.userbenchmark.com/UserRun/1940829

 

I've seen Userbenchmark I think... it was on a website before last I saw it. I disliked it because only the first test it ever ran used SLI; the other two would run single GPU and then claim your system was weak.

 

I haven't investigated how it tests GPUs, but if the window that was open was an indicator it looks like an OpenGL test to me, like Passmark, which is essentially meaningless... especially for mGPU systems. What I think seems unique about it is how it tests separate system components and ranks them against previous submissions. As you can see on the memory test, there are not enough results for DDR3-2400 SO-DIMM comparisons, so it did not rank it against the common DDR3-2133. However, it still analyzed its performance.

CPU@ 4.5GHz - http://www.userbenchmark.com/UserRun/1940968

 

OGL can use SLI without problems, I don't understand why you say it's meaningless?

Yes, this in itself is unique, I'll give it that.

 

I'm not sure OGL can SLI in windowed mode like DX. The test runs in a window, not full screen. For some reason it tests the GPU but does not record the results in the benchmark. Maybe it does not recognize 980M.