EVGA SC17 1080 finally available

The FPU with GPu test only creates a lot more heat right? Compared to selecting cpu, cache together, I mean.

Enviado de meu Pixel 2 usando Tapatalk

 

Cool, I will probably need your help with this one when I visit my Dad in Arizona sometime next year.

Enviado de meu Pixel 2 usando Tapatalk

 

On team LHz you have iunlock, DeeX and zergslayer.
Former LHz member was Mobius1

 

Yeah, expect some more heat. But not on same level as with P95 and the virus Furmark.

 

Actually a good test, because it detected error a lot quicker than the usual with other stress tests to determine a stable voltage. Thanks.
The last two laptops I tested FPU+GPU as you suggested in another thread, would create extreme heat, but both had optimus, so Aida would stress the iGPU and the dGPU together plus the CPU.

Enviado de meu Pixel 2 usando Tapatalk

 

Obviously with this particular torture test, Aida64 FPU+GPU, there is no headroom for overclocking. Liquid Metal would be needed for that.

 

Maybe FinaleWire could make a clone of their Aida64 test tool for Evga? Then could Evga expand their "special" Bios with a stress test tool similar as Dell's ePSA test tool. Maybe they then stopped push out advertising like this below?

 

I suppose an updated vbios is needed to break 23k in FS graphics.
https://www.3dmark.com/3dm/23721420?

 

No kidding, can't even increase the PL limit in the Bios from 60W max, so even if thermals permited, anything over 4,1Ghz would reach the PL quickly and throttle.

 

Yeah, A nice pretty Bios GUI can’t replace the real deal... Like useful power settings. The Overclock advertising is almost as Scam. See... "Yet when we tested both laptops using Handbrake, the Razer Blade Pro was the clear winner, encoding a video file in just 420 seconds. It beat the SC17 1080 even when we converted the file using the SuperClock Mode , which performed the conversion in 429 seconds. Even the Alienware 15 R3, which is based on Intel’s Core i7-7700HQ chip, converted the video file faster ." A fully locked down Intel chips running 3.4GHz on all 4 cores.

 

I am not sure how standardized is that test the reviewer did, but I am doubful of the conclusions of the results, since I ran cinebench like 30 times in a row with 40x (superclock setting) getting 887 consistently, and far from the PL limit. Then I did the same upping it to x41 getting a consistent 908, also not reaching the PL limit. Pretty sure the RBP can't handle a x41 overclock for very long.

 

if you live in antartica it probably could

 

On battery, it will throttle even more. @Papusan place is perfect spot, all BGA testing factory must be moved to Norway and I bet not even one BGA will come out of that factory alive or even flying colors when Papusan tests them.

 

no bga machine should come out of any factory branded as "high performance"

 

All of the Jokebooks would fail and every single OEM had to replaced them with proper LGA + MXM laptops, who contain decent cooling No thin and flimsy would be pushed out after proper testing!! Thin and flimsy notebooks would only come with low power U processors and with iGPU. And none of them would be developed with Nvidia or AMD graphics.

THIS!! +rep

 

Even though I got a 10% discount through my Dad's company, it is a real bummer to see it on sale a month later at $ 1899.99.

 

@Vistar Shook
You said that you tried unlocking the Bios.
Can you please tell me EXACTLY what you did and what and how you got a security error?
Like precisely what you did?

Also, are you able to download EVGA's bios file (is it a BIN file or an EXE file?) and were you able to open it in AMIBCP, and see the Bios "capsule" options that way?
Is this an APTIO V Bios? or some other Bios code?

As far as the heatsinks....I looked at the pictures..are you sure those holes are not "hex screw" type holes? I think I do see one "screw" that looks like it's originating from the other end....It is indeed possible that the screwholes are on the backside of the mainboard, which is going to make repasting a ....biggest pain of almost any laptop, because it would require complete disassembly.

 

Hey @Falkentyne , thanks for stopping by.
I don't have experience modding Bios, so I just did the simple way to see if it would work.
Saved the Bios using afuwin. Opened it with AMIBCP (Aptio V), and I was able to change settings to user. Tried flashing with afuwin the altered bios, but no go. I know this is not the best method, but it did work on another laptop that doesn't have the security thing.

 

I asked EVGA support about the limited Bios not allowing me to change the PL1 limit, not even with XTU, and they answered,"The power limit is definitely by design but there is of course room for improvement but only time will tell if we do decide to continue manufacturing laptops in the future."
Which makes me think that this whole laptop thing might be a short experiment for EVGA, since it is a low margin business with a lot resources dedicated to support, RMA, and such. That would explain their recent sale offering up to 54% on their laptops, to clear stock, when there is not really any new models coming out anytime soon.

 

You can't use AFUWIN.
You need to use FPT64 to flash.

Just remember to set all the bios settings you want unlocked to Supervisor first.

 

Try using FPTW64.exe to flash it. I can't upload it here. Maybe you can find it on the win-raid forums or that Bios Mods forum where dreamonic posts.

 

Try this link

https://1drv.ms/u/s!AkQqWXPLNfg-gWpdLpteUqK2imWT

 

Great, flashing with fptw64 worked, now I just have to figure out how to increase the PL1 limit, all the options I enabled in the Bios don't have an effect. Perhaps it is controled by the EC firmware?

 

What did you change?
I opened your Bios and there are several fields for power limit 1.

I forgot where it was though.

It's NOT under "Config TDP configuration" however, although it won't hurt to change Power Limit 1 and 2 to 200 watts for Config TDP Up, down and nominal (although the number used may be in milliwatts, so in this case, 200 may be 200000 (you have to read the help thing to the right).
However usually config TDP is ignored in many cases if a master power limit setting is in effect elsewhere.

Did you change the PLATFORM POWER LIMIT 1/2 settings?

THESE have NOTHING to do with the 'normal' power limit settings. These are designed to be used as 'extra' throttling features, in conjunction with PSYS (PowerOfSystem) (e.g. PSYS_PMAX). Changing these to 200W won't hurt anything (and might be a good idea in case your cancer firmware is actually trying to use these), but this isn't the target.

The problem is, the "power Limit 1 and 2" should be in the "overclocking" section, where you see all those CPU multiplier settings (and of course EVGA Cancer limits this to 46....I can run mine at x50 but i can only game at x48).

I can't find it but I found it earlier.
It's in a place that says "power Limit 1 override"

If I can't find it now then no wonder you can't find it. It's there though.

I see it in "Bios Strings" but I don't remember WHERE I found it in the APTIO capsule last night. It's pretty well hidden.
The help string says:

Enable/Disable Power Limit 1 override. If this option is disabled, BIOS will program the default values for Power Limit 1 and Power Limit 1 Time Window.

 

@Vistar Shook

OK I found it. How amazing.

View/Config turbo options.

Try setting Max turbo power Limit and Min turbo power limit to supervisor rather than default.
Then change Package TDP limit to supervisor also.

You may need to change these just to get the below settings to work. I dont know how this works but I suspect if they are not set, then the below settings may not do anything.

If you wind up seeing "max turbo power limit 2047W (or 4095W or 1023W) then you know it worked.

Then change power limit 1 and power limit 2 to supervisor.

below that, change power limit 1 OVERRIDE to the same thing.
One of these may be conflicting with the others. there is yet another power limit BELOW that.
the fact is your cancer Bios has 3 entries for Power Limit 1 and 2, all in the same area. And I have no idea which one does what.

 

Pinging the great @Papusan

Having any luck with the Power Limits?
Also wondering if you can exceed 46x multiplier too.

The BIOS supports up to a x83 multiplier.
If you're limited to x46 in the main settings that are not hidden from the user, perhaps you can try the "Overclocking Performance Menu" (which is hidden and has to be unlocked) and go to the Processor area and see if you can go higher there. Even if you can't boot (be careful, I do NOT know how to clear CMOS if you go overboard; x47 would require at least 1350mv just to get the system to POST for sure):

Sets the maximum OC Ratio for the CPU Core and Ring. Uses Mailbox MSR 0x150, cmd 0x10, 0x11. Range 8-83. 0 is AUTO.

 

Indeed I set all the power limits available in the bios to supervisor, and changed them accordingly, but no success. Either way, I was just curious to see if it was possible and perhaps push a bit the limits of this design, but for my needs the 60W limit will do for now, especially due to thermals until a repaste is done. Leaving tomorrow, when I get back, I might mess around a bit more in the settings. Thanks again for all your help, happy to be able to flash the "unlocked" bios and learn new things.

 

You're welcome. How do you know that you are still capped at 60W? Are you using EVGA Windows utilities or Intel XTU that might be overwriting your new settings?
If you happened to have XTU installed, did you set intel XTU to default (activate the default profile) and then uninstall it (IF you were using it?), because XTU is known to automatically revert Bios settings on boot if it was used in the past. This must be done, then after having XTU activate the default profile and then uninstalling it, then you have to go BACK in the Bios and set the settings again (even if they appear to be set; I've seen people mention this on forums in the past).

Is Throttlestop showing a 60W power limit cap? How are you monitoring the power limit?

 

Yes XTU was set to default and uninstalled with Revo, also EVGA Precision Mobile is disabled....I am only using throttlestop and hwinfo.

 

And what are you seeing in TS - TPL?

 

This.
I do find it VERY hard to believe that the EC has a programmed 62W power limit (probably 62, not 60) rather than a Bios default. Because 62W is the default long power limit for 7820HK if using "Default" power limit of 45W (if power limit override are disabled), with 28 seconds burst time allowed, before it reverts back to 45W, and 7700HQ CPU is hard set to 62W power limit for 28 seconds, then reverts back to 45W after, until load stops. So we need to see what's going on. As well as what Throttlestop shows in Limit Reasons, and your TPL settings as well. However it IS possible, because in your default Bios, the power limit overrides were disabled, rather than set to 45W. If EVGA actually set an EC controlled PECI power limit of 60W instead of setting it through the Bios....well...this...is...the ABSOLUTE WORST CANCER I HAVE EVER SEEN IN MY LIFE.

If you are using Throttlestop and you installed it originally fresh, TS may have "captured" the original MMIO power limits and may be set to 60W. In that case, just change Package Power Long and package power short to 200 manually in Throttlestop.

We "may" be able to test this btw.

Go into CPU VR settings, IA Core domain, go to IMON OFFSET, set a negative (-) IMON offset of -31999.
This will cause a CPU to report about 20% lower power consumption than it's actually really using. Then if Throttlestop shows PL1 throttling at something weird like 48W rather than 60W, then we know the EC is doing something to you. (if you're able to use "more" power than before, with the power reported lower than 60W, with no throttling under 60W, then you at least gain something).

 

Sorry guys, I am finally back at home and have more time to play with the machine. Since messing around with the Bios didn't get me anywhere, I just assumed that the power limits are not controlled by the Bios settings. I will post later the throttlestop tpl and limit reasons.
I still think it is EC controlled, because if I select the normal profile in Bios, it will have the standard 45 PL1 limit and 56W PL2 limit. If I select superclock or custom profile, then PL1 is 60W. If I unplug the laptop, on battery it will be 40W on all profiles.

 

Due to the TDP limits of the CPU and GPU, and of course the thermal limits of the design, this is the best I can get in Firestrike without modding.
graphics 23304 and physics 14057
https://www.3dmark.com/3dm/24233352?

 

Since I flashed the new Bios from EVGA to patch the ME vulnerability I will flash the modded Bios later and see what results I get with the IMON OFFSET test.

Thanks again for the tips.

 

I forgot to give the IMON OFFSET a go.....and guess what, it works! Now of course I need a repaste to handle the thermals.
Thanks man, awesome help.

 

My CPU doesn't undervolt very well, so at 4.4GHZ I need a lot more voltage than what others get with this chip......silicon lottery. Let's see what LM can do.

 

Cinebench score before was 929 before power throttling.
Now I managed 980 with the imon offset. Can't really go any higher due to thermal throttling at higher clocks on stock paste.
Edit: this score is 4.4GHz not 4.3....sorry for the confusion. The point is that at least with the Imon offset, it can surpass the 60W power limit.

 

Nice score for 4.3GHz

 

Wow you beat my score. I get 970 at 4.3 ghz. And that's with RAM timings at 15-15-15-35, 1T, tRFC=270, tREFI=32767....
and @Papusan is actually happy with your score...what...ive never seen him happy before.....BGA jokebook with a good score.......

 

I FIX IT FOR J00
If it was from 4.3GHz. But I doubt it

 

I wish....at 4.4 .
Sorry I edited the cinebench post with the higher score but didn't edit the post before mentioning 4.3. Sorry for the confusion.
At 4.4 I did reach 980.

Enviado de meu Pixel 2 usando Tapatalk

 

Yeah, as I expected But you beat the AWbook's running 4.4GHz +Rep already posted.

 

Cinebench score did go down a bit.

 

What is down a bit? Numbers

 

@4.4, I am getting 975 instead of 980, so 0,5%. Haven't tested other benchmarks though, so far within margin of error with cinebench.

Enviado de meu Pixel 2 usando Tapatalk

 

How did EVGA achieve this? Mine is still susceptible to spectre.

Change process priority of CB R15 to Realtime in Task Mgr's Details section

 

ran it again, and well within margin of error.

 

Does changing the process to realtime make a difference? I will give it try. Thanks.

 

That's how my BGA chip scored 696 on cb r15 when it scored only 678 most of the time.

 

Well, renever I try to run it on realtime, it freezes my machine, even on stock clocks and voltages. Setting it on high as well. AV and unwanted services disabled as well.