Clevo Overclocker's Lounge

There are already rumors that time spy used a different version of Async Compute more or less tailored to show Pascal's strength with pre-emption. It uses something called concurrent compute instead.

 

Might actually be true, specially if you look at how 1070 scores. For the parts where the 980 OC'd to hell barely crosses 20fps the 1070 pushes out 45fps at stock.

 

That was the basis of my previous sarcasm. Why am I not surprised? The NVIDIOTS have no morals. They use cancer drivers and firmware AIDS to cripple yesterday's flagship GPUs in order to get the job done where their hardware falls short standing on its own merits.

Ah, now there you go again... using your brain. Logic doesn't apply any more. The world is dominated by stupid now.

 

LOL.

btw. i did use the latest 368.81 @j95 drivers for that run. Seem to be working great.

 

if I remember correctly in time spy

Essentially maxwell gets no improvement with async on or off, pascal gets like 7%, AMD gets like 10%. Maxwell's performance is probably due to limitations nvidia cant fix via driver. I am not farmilar with this issue at all so I could be completely wrong. I heard also that if real async is used(providing the rumor is true) maxwell could see a performance decrease.

I would bench my 1080 but that means I need to get windows 10.

 

Cheap...
https://www.kinguin.net/category/19429/windows-10-professional-oem-key/

 

Where are you getting these scores from. On stock? 1070? Sound like the card I need to go find and bench.
http://www.3dmark.com/compare/spy/64912/spy/72267/spy/22752/spy/26065/spy/53431/spy/22518
I gotta say. I don't see that 45+ on stock???

 

Oh nono, sorry i meant in "sections" of the Graphics Tests where the 980 struggles to manage 20ish fps the 1070 easily hit 45fps.

For example :

 

To be fair, so does the RX480. And it would struggle far more if they ran the bench at the right resolution.

 

True. RX480 im happy about, the 1070 not so much. Specially with NVIDIA and the Microsoft DX12 shenanigans.



Also does anyone know what the bandwidth / lanes are like for the M.2 slots on the Phoenix?

Mainly is it the same for the 2 SSD slots and the WLAN card slot? Or it is different for both?

The reason why im looking for this info, is because im planning on ordering the PE4C v4.1 to test out what eGPU solutions will work with this baby, if it works well and good if it doesn't ill just sell the adapter.
Need to order the right one for the slot with the max bandwidth.

PE4C-M2060A V4.1 or the PE4C-M4060A V4.1

 

I would be happy with the RX480 if it came out when it should have to compete against the GTX 980, but it didn't. So can't really get excited about that.....

And yes they are still with the shenanigans and DX anything these days. all one can do is ride it out and hope at some point in time we get the better end of the deal. And that wont happen till AMD gets current with Nvidia current top of the line gpus and not yesteryear ones. IMHO

You would need the m.2 key M. The slot has to be on the right when looking directly at the connector. On the left is the normal m.2 slot or m.2 key a/e

I think.

 

Turns out the system has 2 x M and one A/E, slots.
Thinking of going with the M version, since i rarely use the second one with the 850 Evo in it and always keep on swapping around the 850 Evo with a second 950 Pro.

Spoiler: pikchurs

 

Man I wish this had those slots, but even adding one in it wouldn't be the M slot, but the A/E which is far slower......


If you can route the cable, im 100 percent sure it will work. gong to stay positive here.

 

Yeah routing isn't an issue, my keyboard isn't screwed in, and i rarely use the internal display, just going to route the cable from under the keyboard and right out in front of the display

 

Sounds like a plan...When to you expect to have this plan operational? I was thinking of doing the expresscard to hdmi in and expresscard to external video card mod, but haven't gotten around to that yet.

 

Well the PE4C V4.1 wont be here before the 30th at least. Hasn't shipped out yet. But before that gets here ill get to try the Razer Core with my system. If that works, this might become moot since, i think, over TB3 the bandwidth is much higher.

 

Time spy DX12 benchmark on my 1080 FTW overclock.
http://www.3dmark.com/spy/85669

IDK why it keeps complaining the time measurement data issue.

 

Probably doesn't like the clocks. Just move the CPU frequency a bit. (a few mhz)

 

This bench isn't finished. More like a beta or drivers don't play well with it!! I expected more from Futuremark.

 

Maybe, but i faced the exact same issue with FireStrike and i did what i wrote in the previous post. (was suggested by @johnksss )

 

Or latest sys info version from Futuremark sucks. I'm still on holiday and haven't tested this ****.

 

@bloodhawk - any reason you are wanting to use an M.2 slot instead of the Thunderbolt 3 port? It should run at 40GB/s and use PCIe Gen 3, should it not? Seems like that would be a much easier plug-and-play option. Just curious if there is some kind of drawback that justifies using one of your M.2 slots instead of the easily accessible TB3 port.

 

Well with the latest TB 3 firmware update, Clevo locked out eGPU support. I have been talking to Prema about this, and he has been working on something for TB3 as well, just hasn't gotten enough time. Prema might be able to unlock it, or might not be able to, he isn't sure. But till that works out, i just wanted to play around and see what works and what doesn't, maybe the data will be able to help him out in some way ?
I do have a friend bringing he Razer Core over around the 29th, so we will know then if TB3 eGPU's still work or not.

 

The wifi card is single lane IIRC.

 

http://www.3dmark.com/3dm/13377795

 

wth would clevo do that? maybe thinking about bringing their own proprietary egpu solution to the market?

Sent from my Huawei Mate 8 NXT-AL10

 

"turbo boost short power short power max" is greyed out on my XTU. is there a way to enable this? is it safe to do the basic clocks as noted here http://forum.notebookreview.com/threads/clevo-p870dm-n00bies-guide.789463/ even if i can't change that setting?

 

hmmm....crysis 3 crashed my GPU so sticking with stock clocks.

 

That and to protect their high end systems. And MXM based profits. Seems logical tbh.

At what clocks?

Nice!

What did you change? Also were you able to figure out what the cause was?

 

So after checking my bios its different now from the prema bios i had before i they(helix) sent my mobo to sager. My boot logo is back to sager instead of the prema logo i had also.
1st pic is before my system broke down

Can anyone help me on getting back to 4.2ghz? Or should i just reflash my system with the previous prema bios i had. @Phoenix how do i apply the old instructiob/tutorial you sent me when some settings is not found anymore.

 

That looks like stock BIOS.

 

Still trying to confirm precisely what made things better since I do frequent clean installs. This time I installed from a heavily modded WIM that had all of the Windows 10 poo-poo components (Cortana, telemetry, modern apps, modern app associations, Windows Store, etc.) exorcised using the MSMG Toolkit as explained here and I did not enable HPET or unpark CPU cores. I also increased my CPU core and ring voltage settings in the BIOS from 29, 29, 29 to 40, 40, 40. I also did not activate Windows or connect to the internet until after I finished installing all of my own drivers and then disabled Windows Update to block Micr0$loth from installing any of their own cancer drivers. Oddly enough, the CPU is not throttling with the higher voltage and the load temps did not increase. I am going to install a few more of my basic software and create an image and then start changing things one at a time to my normal setup configuration to see if instability surfaces with Time Spy so I can identify what the offending tweak is. I have a hunch it was CPU voltage too low, HPET being enabled, or a combination of both.

 

I dont think the stock had limit things. Hold on my upload speed is bad but i am currently uploading a vid of my system as it boots up

 

Well, there is a good chance they just flashed a slightly unlocked version of the stock BIOS. I would just flash over the Prema BIOS. Just make sure the EC versions are the same.

 

How do i check if the ec versions are the same?

Sent from my SM-G935F using Tapatalk

 

That is such BS that the benchmark performs better without any tweaks. Feels like they are trying to even the playing field or what not.
But keep us updated, ill try to run a few test tonight as well.

Should be on the first page. And If you got your Prema BIOS from Prema directly, you can ask him.

 

I agree. Something fishy going on.

 

Here is the video of the boot up . Clearly it says prema but idk what happened. @Prema here is my ec version
will the bios work with this? Weird as i only asked them to repair my socket i dont understand why this happened.

Sent from my SM-G935F using Tapatalk

 

HaHa. Everything is possible if M$ is involved in this benchmark I bet they have put a lot of $ in this!!

Edit. And the site functioning again

 

That is weeeeird.
Can you post a few pictures of the Overclocking menu ?

 

i could not find the ring settings etc. Tons of things are missing @Phoenix @Prema

 

I cannot resist the temptation to post this here because it's pretty stinking awesome. For those that haven't seen it @Khenglish had a little mishap with one of my 980M cards (details in opening post) and, long story short, instead of scrapping the GPU he transplanted a desktop GTX 980 desktop GPU core on a 980M MXM PCB and it was a success... getting some pretty awesome benchmark scores. His talent is truly amazing.

https://www.techinferno.com/index.p...0-core-upgrade/&do=findComment&comment=146309

 

Then they simply flashed the stock Bios over it and some remnants of Prema's Bios seems to be still there.

So long story short: either you flash back Prema's Bios or send this unit back for a refund....

 

Ill be flashing prema bios maybe before the week ends if i find some time.

Sent from my SM-G935F using Tapatalk

 

@bloodhawk - definitely more CPU voltage helped. Added an extra +10mV offset to keep it from throwing the stupid Futuremark time measurement error flag. So, +40mV non-turbo, +40mV turbo and +50mV offset. (4.7GHz is stable with 29, 29, 29.)

Single 200W 980 at 1500 core 1965 memory, 6700K @ 4.8GHz. System pulled a peak of 462W on my UPS digital meter with this OC.

http://www.3dmark.com/spy/95134



Might as well toss this in here also, since I used it to confirm my CPU overclock was dialed in for Time Spy.

 

Nice to see Time Spy over 5000 Is it test one that draws most wattage in the bench? And +460W with 1500/1965 is pretty hefty. Imagine what 980 Sli would use with Max Oc What about max cpu/gpu temp?

 

Yeah, 980 SLI would probably be up over 700W at similar clocks. It doesn't seem to draw more than 3DMark 11 or Fire Strike. Very close to the same. I think it's just a buggy program more than anything else.

I am benching in on AC, so nice and cold. Like 23°C on GPU and 62°C on CPU.

 

Nice!Amazing score!!

Ill give the bench a shot later tonight!

 

If you can max gpu in bench with 23 degrees.... I would buy similar AC and bench next winter Firestrike is still weak against Sky Diver. Can Time Spy force more Cpu power?

 

I have not measured that. It takes more voltage to do the same OC, so either it takes more CPU power or the benchmark is just screwed up. Not sure which one.