Khenglish

What the P6xxRx systems likely do is power off the dgpu when the igpu is driving the internal lcd. Notice that Prema also said "no optimus". It sounds like the igpu mode is intended only for battery life extension purposes, and the external display ports will simply not work at all in this mode.

7 hours ago, AtoningUnifex said:

@Khenglish: are you saying this about all laptops? Or just the Clevo range I was asking about?

 

It is true that most laptops these days only allow a single GPU to be connected to the outputs at a time, but this isn't necessarily true of all laptops, particularly those with mux chips. Before Optimus became common, some laptop models used mux chips to allow both GPUs to be running at the same time with different outputs (although which outputs could be used with which GPUs varied from model to model). This appears to be rare these days, but that doesn't mean NO laptops do this anymore.

 

According to a post by @Prema (apparently before release):

 

Can someone confirm this is the case?

 

And if so, when the internal display panel is connected to the iGPU, is the dGPU turned off and the external ports non-functional? Or can the dGPU be enabled given the external ports are connected to it?

All laptops with a mux select either all outputs as dGPU, or all outputs as iGPU. While it is possible for the manufacturer to easily expand this to pick and choose which GPU does which, no one ever implemented doing so. This is true for all Clevo, MSI, and Alienware systems. Thinking back, during the Sandy Bridge era there may have been a system or two that had the dGPU only wired to the displayport, but that was the only exception (M18x R1 and M17x R3). I might not even be right about that, but I vaguely remember people saying their DP port only worked in PEG mode.

You can't use both the iGPU and dGPU simultaneously on laptops to drive displays. You are either stuck with iGPU, stuck with dGPU, or can choose between 100% iGPU or 100% dGPU. You can't use both to drive all display outputs at once.

You want the setting "Enable applications for optimus". Try setting it to "enable".

If you have access to the Nvidia control panel that means the driver is running and you are not having hardware issues. If you were the driver would not run.

Download Nvidia Inspector. There is an option called "optimus shim" or something like that that you must make sure is enabled. If you can't find the option I will make a screenshot.

I can't get win10 to not run the frame rate limiter, so I am using win7.

Can you open the Nvidia Control panel if you right click on the desktop?

32 minutes ago, ajc9988 said:

You get bonus points if you can throw a 1070 into a P700 series (the 980 and 1080 were never options with current cooling and size, unless doing what Khenglish did and throwing a 980 core on a 980m card, which may not work with the 10 series)...

Sent from my SM-G900P using Tapatalk
 

Unfortunately I don't see any good BGA swap options for this series. 1080 core cannot go on 1070 due to GDDR5/GDDR5X difference. Mobile 1070 core has more shaders than desktop 1070, so no upgrade there either.

Maxwell and Pascal BGAs are much different, so no go on putting a Pascal on a Maxwell board.

The MSI 1070 can likely be crammed into systems with relatively minor mods. The MSI 1080 would require extreme case mods, heatsink mods, and some motherboard mods, but if I somehow got one I would definitely do them.

Usually when that happens the battery is just too old. Do the following to try to recalibrate:

In the power options set it to sleep at the critical battery %. Drain the battery until this forced sleep, then immediately turn the computer back on. Then use the laptop until it shuts down without warning. After this then let the laptop charge back to 100%. If there is no battery life increase after this, you need a new battery.

5 hours ago, Prema said:

Added 'clean' shots of CLEVO 1060 to the OP!

 

@Khenglish MSI and CLEVO 1080 MP boards share identical thickness.

Out of curiosity is this the standard pcb thickness, or thicker like the 7970m and QS kepler boards?

1 hour ago, Prema said:

OK, added 'clean' MSI GTX 1080 (single) shots to the OP!

 

@Khenglish What does our 'GPU DOC' have to say about Clevo and MSI 1080 designs? 

Honestly both are overkill and majorly beat out the 1080 Founder's edition. The voltage regulation on either card is good enough for LN2 overclocking, The Clevo is rated for 480A on the core and the MSI is rated for 640A.  The Desktop FE can only do 250A, and that's questionable as it relies on its FET's pulse current specification.

It looks like the MSI only has one memory phase while the Clevo has 2, but the MSI has more core phases. I'm not sure of the core phase configuration on the MSI, but it looks like an unheard of triple VRM design, with 4+2+2 phases on the core vs the 4+2 on the Clevo. Honestly I would consider any thing more than 4 phases to be unnecessary, and both greatly exceed 4 phases. The Clevo will likely be a little more power efficient as the lower 2 core phases pull juice from the very close by MXM power connector which can supply a very high amount of juice, while the MSI will pull relatively more current from the power cable.

An unknown at this point is the PCB thickness. There are 2 thickness options to still fit in an MXM slot, and most cards were the thinner option. If one card is thicker than the other, that would outweigh any other difference between them. You'll get much higher power efficiency and likely memory clocks too.

I was hoping MSI would match the 1070 form factor. It looks like the MSI 1070 is the only Pascal card that can be fit into an existing MXM system without ridiculous mods. The mods may still be rather extreme (it looks like I would need to remove the subwoofer and slide the SATA slot slightly forward on the P150EM motherboard), but at least the core placement is the same as standard MXM 3 cards and the GPU fan is not in the way.

The dorms across the street from my apartment just got their wifi upgraded, and I would love to ditch TWC in favor of the far faster and far far far far more reliable college wifi that my tuition already pays for. My  laptop can reach it, but unfortunately due to the range the wifi does occasionally get dropped (although better than the TWC drop rate), and phones get far too weak of a signal to use the wifi at all.

I was wondering if there was a wifi repeater with very good range that could rebroadcast the college wifi within the apartment. The idea is I use the repeater to log in to the wifi, and then the repeater rebroadcasts with a new name and password that us in the apartment can log in to. Unfortunately every wifi repeater I find will only broadcast a unique wifi name and password if it receives an ethernet connection, not a wifi connection. We cannot rebroadcast as the same network as that will let too many people be able to log in and network admins would likely be mad at me. Is there a repeater than will take and rebroadcast a wifi under a different name with range as good as a laptop?

Unfortunately TWC has a monopoly in my area, so grabbing the college wifi is the only way to get decent internet.

2 hours ago, Prema said:

http://www.3dmark.com/fs/9794614

From what I've read everything beyond a 1080 is cpu limited in firestike regular. How does it do at ultra?

While I do like higher refresh rates, I like higher screen res even more. I plugged my laptop into a 2560x1600 screen at work and loved the extra detail despite the lower refresh rate and worse response times. The currently plan is to route my external DP port internally, and run a 2880x1620 display. 4 year old ivy bridge is still competitive, so wait for kaby lake and zen before considering updating the entire platform. If I can somehow snag a cheap MSI 1080 I'll do that, but otherwise no more GPU upgrades on the P150em.

13 hours ago, Hwrgrabe said:

 

Thank you Khenglish for replying. 

 

So I went back into grub and set to what you tried. 

 

0x16f to 0x01

0x170 to 0x1d

0x171 to 0x00

 

My max GPU speed is still 1250 mhz as shown by GPUz utility. The only thing I have done is edit these UEFI variables. Is there something else that I need to do? The normal max has always been 1250 mhz. 

 

Thank you once again for reading.

The issue is likely that you aren't running an xm cpu.

Also I found that the frequency is always 50MHz under what is expected for the multiplier. When dropping it by 1 I then got 1400.

13 hours ago, Hwrgrabe said:

Sorry to alter the discussion, I would like to refer back to the iGPU overclock enabler issue I had and still have. I have my UEFI variables set as follows:

 

0x16f to 0x01

0x170 to 0x1f

0x171 to 0x25

 

I used GPUz to measure the iGPU speed and ran a stress test, but the max speed of 1250mhz has not changed. Is there something obvious that I'm forgetting to do?

 

I also have an additional question about setting up an E-GPU for my latitude. Will the following parts be sufficient or will I have to buy other cables/adaptors?

1. Bplus Pe4c PCIe 16x to Expresscard (<-linked): Is this compatitble with the EC slot in the latitude?

2. EVGA GeForce GTX 750Ti (<-linked)

3. Dell DA-2 220 W (<- linked): Do I even need this? I heard that the 750 Ti can run without a external PSU?

 

This list seems rather short and thus I am skeptical. Additionally, has anyone tried EGPU with this model before (I would like to PM you)? Will I just need to download Nividia Drivers or will I have to go the Setup 1.3 route?

 

Thank you for reading so much, I appreciate all the help you guys have given me. Someday I'll be the guy helping out. 

 

 

I just tried:

0x16f to 0x01

0x170 to 0x1d

no extra voltage

This should set 1.5ghz, but instead I got 1.45GHz. Still a 100MHz overclock for a 3920xm. iGPU current limit may be a limiting factor.

41 minutes ago, Meaker said:

It's funny how clevo ships the 1080 (with GDDR5x) with the stronger memory controller phase layout (160 amps lol) over the 1070 which has (the more power hungry) GDDR5.

That's not memory. Those are 2 core phases. The core on the 1080 Clevo cards is 4+2. I don't get why the 2 phases have a cut down FET count on the 1070 while the 4 are unchanged. Seems like the thing to do would be to cut both, or cut neither.

The memory is the two phases with the 2 smaller .24mH inductors.

22 minutes ago, Meaker said:

http://www.ti.com/lit/ds/symlink/csd87350q5d.pdf

 

VRMs on our 1080 PCB.

That's the same chip used since the 680m.

Its also on some desktop cards and motherboards.

1 hour ago, Prema said:

 

Current standard (since 2012) is MXM III (revision 3.1, which added PCIe 3.0).

Type 'b' (and 'a') simply describe the physical dimensions, hence the new cards are MXM III (3.1) but not 'b' (apart from the MSI 1060).

The original MXM III from 2009 can still run a modern 980m. Only pci-e 2.0, but still that's 7 years of upgrades. For old laptops like the m15x the rest of the laptop was long obsolete before it ran out of GPU upgrade options.

@mcdudehouse

The dGPU always reverts to 202MHz and sticks there when using an unrecognized PSU. I found no way around this. The CPU runs stupidly slow too, but that's correctable with Throttlestop.

1 hour ago, ajc9988 said:

If only the industry would settle on a similar standard on shape like mxm did and create an aftermarket of both heatsinks and cards like desktops! That is where the industry needs to move to! Standard MB and component placement, heatsink, and videocard. Maybe have extended space on certain laptop models advertised as such for oversized cards like you would have to pick a case for a desktop to hold the oversized video cards. A man can dream of a day for just buying a "case" (chassis), display, mb, cpu (desktop chips for all, allowing for low power desktop chips (which now have tdp matching mobile sockets, or full bore chips), gpu, heatsink, ram, etc. And building it from scratch, upgrading as desired, buying his preferred manufacturer on each component. This is where modular laptops need to go! Then laptop manufacturers would also need to compete on quality of each component and you would have high end, highly customized machines! :-)

Sent from my SM-G900P using Tapatalk

While its nice that laptops are getting gpu hardware as close to desktop stuff as ever, I overall am not liking this change in laptop cards starting with the 980 notebook. Upgrading seems dead. While standard mxm cards were overpriced, the cost of these custom mxm cards is just nuts. Prices for the 980 notebook cards were over $1200, and I don't expect the 1080 to be any less than that. We used to have Dell, MSI, and Clevo all making standard and mostly interchangeable cards. Now Clevo and MSI are incompatible and Dell seems to have disappeared.

I'd rather have slightly weaker, but much cheaper and upgradable GPUs than slightly stronger and much more expensive and unupgradable GPUs. Right now they still use the old standard mxm slot so some laptops can probably use new cards with modified laptop cases and heatsinks, but I expect that to be dropped next gen fully killing gpu upgrades.

It would be great is MSI and Clevo got together and aggreed on a mobile GPU standard that they would try to continue for several generations.

2 hours ago, sirana said:

And if the single 1080 is running in a P775DM3, it only comes with a single 330W adapter. Won't it lack power then? Sorry to bother you so much, I am just anxious to spend my money "wisely" 

With a 980 and a single 330W PSU I need to push the GPU power draw over 220W to have problems. I believe the TDP of the 1080m (yes I am calling it "m" for mobile. this is the laptop version and the "m" always stood for mobile...) is 125W, so actual power draw will be around 150W. It should be ok at default, mild, and maybe moderate overclocks, but expect PSU shutdowns if you start raising the voltage much.

20 minutes ago, ajc9988 said:

Most likely heat. Check out the p775DM3 heatsink and compare that to the new 750dm2! The extra heat offload with the extra pipes seem to be part of it, potentially... Either that or electrical specs...

Sent from my SM-G900P using Tapatalk

The middle 8mm pipes both CPU and GPU side of the 775DM3 only cover half the radiator.

It is nice to see clevo to start to use big 8mm pipes at least.

Any idea why the 1070 is available for the P750DM2, but not the 1080? Will the 1080 be available at some point?

Also are the 15.6" models getting 120Hz screens, or just the 17.3s?

Prema any chance you could fix the MSI 1080 links? Is the layout the same as the 1070? The MSI 1070 looks really good. It should be fittable into a standard MXM 3 slot and be able to take an unmodified heatsink. Too bad MSI cards can only be found rarely on ebay.

I am very surprised by the use of multiple core VRMs. The only cards I am aware of that being done on were pre 2011 AMD top end GPUs (HD 4890 and I think 2900XT). The Clevo is definitely 4+2. The MSI is hard to tell. I first thought it was a 5, but it looks like it is actually a 3+2. There are two sizable VRMs at the top that look like they are for the core, while neither is big enough to be able to control 5 phases. I feel bad for Prema trying to BIOS mod these multi-VRM cards.

The Clevo design does look a little better, but of course its bigger. Those lower 2 core phases will pull power very efficiently from the MXM power tabs, and the upper 4 phases get power efficiently from the power connector. Also 2 memory phases should outperform the single memory phase on the MSI. The MSI appears to be plug & play into older laptops though... well worth the less than 1% overclocking difference.

1 hour ago, Clyde said:

You can give even 16mm square electrical wire and further will not result. the weakest link in P150xM is plug/connector. If you do not replace it forget to the high OC. All my P150 are replaced connector/mobo from P170. BTW, My QS K5000M (9300 points in 3DMark) had poor ASIC but thicker PCB than standard OEM card, so better contact.

When using the standard p150em psu cable on a 330w psu I was getting .4v dropped by the cable, and .2v dropped by the connector. The 330w psu cable is twice as thick as the 180w cable. When I switched the 180w cable with the 330w cable while keeping the 180w connector I gained about .2v. You likely gained another .1v by also switching the connector.

I miss the thicker pcb on the Clevo 7970m. The pcb and components on that card were the highest quality I have ever seen on an mxm card, and it outclocked its desktop equivalent, the 7870. Too bad the maxwell cards did not have the higher end pcb.