incendery

enough benching for one night 😄 https://www.3dmark.com/3dm/116378299

13k is so close! https://www.3dmark.com/3dm/116377044?

More tuning, more tweaking hitting the edge of stability at +185 on the core, tempted to add some more bulk capacitance on the card to see if I can eek out a little more frequency. (don't worry about the memory I've got better stuff on the way 😁) http://www.3dmark.com/3dm/116369231?

good on you 🙂 always a bit nerve wracking doing it for the first time seconding what kojack said with the slippery slope 😂

it's a bit of an oddball solution but perhaps something like this could work? https://www.aliexpress.com/i/1005004111527529.html

I've posted a little bit about this laptop process but thought it would be best to make a thread to document it here. current specs: P775dm3-g i7 8700k: 4.7ghz -115mv RTX 3080 16GB (180w shunt modification) 32gb CL18 spectek 2x16 2tb 970 evo + 2tb WD Blue SSD 1080p 60hz panel intel AC8265 Modified Dsanke "P7xxDM_RTX" 1.07 bios to suit Windows 11 The laptop I started with is a P775DM3-g I picked up for cheap with a 6700k and an unhappy 1060 that would constantly crash and stop outputting to the display. After finding this thread from a forum member who had bought a 3080mxm for their P775tm1 and documented the process to get it working (and the possibility of making nvidia drivers work without .inf modification which was a priority for me) I decided to order a 3080 for myself. At the time I had a QTJ2 mutant cpu but unfortunately due to the IHS height and the rigidity of the aliexpress customized heatsink I had to go back to the 6700k, This was then upgraded to a delidded 8700k acquired from a friend. Unfortunately the 3080 was mildly damaged in shipping! A cracked inductor and the external power connector was squished. in a stroke of luck ebay was helpful with the refund process and I was left with the card in my posession, the connector was an easy fix but the inductor required a bit more effort. With the parts installed I then worked on the bios going off the details in this comment from forum user srs2236, changed the ID from 7705 (p775) to 7715 (X170KM). Installing the latest Nvidia driver worked without a hitch, power was limited to 150w due to no dynamic boost support (power limit is covered later on), the aliexpress heatsink is amazing at keeping temperatures in check. With the GPU working fine I began on the CPU, due to some mistakes with mounting pressure and the QTJ2 I had damaged / squished one corner of the socket so needed to lift a section of pins back up by hand (thought for a moment I might have killed the board when I realized what had happened😐) The P775 has a lower OCP limit than the TM1 (144A vs 180A) so if too high of an instantanous load (avx torture test etc) is applied it can cause shutdowns with a big overclock, I found OCCT small extreme to be too much unless I limit the CPU. currently I have settled at 4.7ghz -115mv offset and it is very stable, photo attached is from testing at 4.8ghz. After some benchmarking and tuning I was annoyed with the 150W power limit but I wasn't prepared to use the old driver with unlocked power limits, so the only real choice was a shunt modification... 20mOhm shunt resistors added on to the existing 5mOhm, in theory this should give 20% extra headroom (180w) but until I can tap in to i2c on the card and see what power actually is getting used I can only make an educated guess how much power it is drawing. I went from a bit over 12k in timespy graphics score to over 14k! Next post I will cover some of the problems and solutions I went through getting the bios to work (I am firmly a novice in this area so a lot of it is trial and error)

I have a 2021 modified version already :), I was playing around with the later version with the bios features. the only small realtek hiccup is no automatic lowering of volume to 30% when plugging in the headphones so care is needed to not deafen yourself ---Update, P775dm 2/3 audio fix--- you can fix the driver issue and use unmodified P7xxdm2/3 audio drivers by changing the intel audio device subsystem ID in the boardID module, this is causing a mismatch with the dsanke RTX bios which stops the driver working. search for: 8B 0E C7 04 08 58 15 09 77 EB 0D 0F B6 45 FF 8B change to: 8B 0E C7 04 08 58 15 05 77 EB 0D 0F B6 45 FF 8B and you will have working audio drivers again. *speaker and headphone audio are not separate, be careful not to deafen yourself when switching to headphones!

Vbios is the ROM that initializes the GPU when you turn the computer on :), it's either stored in a separate chip from the normal bios or it can be included in the main bios ROM. https://en.wikipedia.org/wiki/Video_BIOS some good information to get context here if you can load up GPUZ and screenshot the main windows that would be great Ignore this part I have just seen your new post* Cheers

another vouch for PTM7950, I'm amazed how well it performs.

Hello P7xxDM2/3 users reading this thread, attached is a lightly tweaked Dsanke bios for these laptops with support added for 3000 cards from the X170KM without .inf modification. The only changes from the original P7xxDMx_RTX.rom is a change to GPU subsystem 7709 (P775tm1) to 7715 (x170km) You will need to use the Realtek UAD driver once installed due to a change from HDA to UAD (according to dsanke in the smxdiy post containing this bios), attached is the version of the bios from the p775tm1 that works with the dm3 soundcard, the only issues are improper support for volume control, you will have the same volume set on headphones and speakers but the amplifier will enable and work https://downloads.eurocom.com/support/drivers/systems/431/431_Audio_W1064.zip I am unsure exactly how Dsanke has worked their magic but it is based on the P775TM1 and has extra features in the bios plus support for proper overclocking controls for >4 cores (does not support 6/7th gen CPUs on this bios). P7xx_Dsanke_1-07-24_RTX3080.7z

hmm strange issue, looking at the block diagram the mux only handles the internal display connection. this would explain why you don't see advanced startup grub etc if they're defaulting to the igpu which doesn't have a way to output to the external monitors. I'm unsure why the DGPU doesn't wake up in windows that is quite odd, it may be something to do with the vbios perhaps? there is 2 roms onboard for optimus and dgpu only behaviour and perhaps there is an issue with the dgpu ROM? In the details for microsoft display adapter what is the hardware ID reported if any?

Meaker you didn't suffer from substrate bending did you? I was having this issue with the aftermarket heatsink and stock mounting screws, I replaced them with springs on longer screws and repaired the squished pins in the socket and it's happy now.

hmmm I need to redo my changes to boardID in the bios, inadvertently changed the subsystem ID for the audio device and now my headphones don't work 😂 shunt mod is still working very smoothly, I will push the memory a little now and see what scores I can hit (update) fixed! had one of the wrong values in the boardID file changed, it now works correctly and smoothly

The benchmarking is getting quite addictive now 😄 maximum GPU temperature of 72c even with the shunt mod and in theory near constant 170-180w power draw during the bench, I am interested in getting access to the i2c bus for the GPU to see what my real power draw is. Not quite stable, getting some artifacting on the screen with this run so I will have to up the voltage a little bit, but valid score regardless. Very impressed with PTM7950

Clevo P775dm3 with a 3080 and an 8700k (and some mediocre spectek memory) 3080 is 180w power limited, 8700k is 4.7, had some post issues recently related to socket pressure on the CPU which has been a bit of a pain. +200 memory on the GPU and VF curve holds around 1890mhz And timespy, a few artifacts visible so it's not quite stable but a valid run regardless, very impressed with PTM7950 on direct die

Firetrike score was 35631 GPU after PTM installation, I'm able to hold around steady boost at 1890mhz under firestrike load while under 60c! http://www.3dmark.com/3dm/115925832?key=akYYmlKjjF6uLFjhIQNyfA I do a fair bit of gaming at 1080p on my desktop, I'm moving countries in the next month so this laptop will become my main machine very soon. Should tide me over for a year or so until I rebuild a desktop :)

it seems like the TDP is 115 on the 2070 super according to TPU you may want to look into omgnvflash and if there is any bios modification possible on turing now 🙂 otherwise there is always modifying the shunts to trick the card into using more power.. https://www.techpowerup.com/gpu-specs/geforce-rtx-2070-super-mobile.c3514

with a little undervolting I am quite happy with this score despite being on the raw edge of throttling, holding at 1800mhz core through the whole bench

Clevo P755DM-3 G i7 8700k @4.7ghz 135w power limit RTX 3080 GA104 16GB MXM (Shunt modded ~180w PL) stable around 1800 core on the gpu but at the limit of throttling, will see if ptm7950 improves the temps

I have successfully modified the shunt resistors on a clevo 3080 card out of an x170km, I installed 20mohm resistors on top of the existing 5mohm for somewhere in the neighbourhood of a 4mohm result (in theory a 20% increase in headroom" due to lacking dynamic boost support in my p775dm3 I was stuck at the 150w limit, it should now be around 180w! I am temperature limited now but will update with performance once my ptm7950 arrives! 3dmark timespy result: 13801 GPU score https://www.3dmark.com/3dm/115871921

Tragic to hear about your laptop 😞 The new desktop sounds like a worthy replacement at least

what modifications did it require to get the 4080 working? 😮 inf mod I assume at minimum? great works as always!

I'm looking into modifying the OCP setting on my P775DM3 to stop it shutting down when I stress the cpu too hard 😁 (8700k pulls a little too much power under heavy stress tests) from what I can tell between the P775tm1 schematic and the dm3 they are practically identical when it comes to the VRM and power delivery components so I'm not worried about overloading the power components, the practical differences between the two boards in the schematics seem to be a few configuration resistors for OCP and ICMAX configuration. [P775DM3 schematic page 108, vcore] PR27 is setting OCP here and is different to the tm1 where it's 38.3k I am quite comfortable with SMD rework and have a reasonable amount of experience in this area, does this seem like a sane idea to others or is there maybe something big I have missed in this plan?

Purchased some LM and a sheet of PTM7950, very interested to see the performance of the PTM vs paste. LM for an 8700k delidded using a 3d printed model https://www.thingiverse.com/thing:1630652, it seemed to work well on a dead i5 8500 I had sitting around 🙂 hopefully the 8700k survives

love the Bluetooth upgrade very nicely done 🙂