HP Omen 17 2019 RTX 2080 VBios Modding (190W)

Maybe your high 800MHz VRAM overclock is sucking power from the GPU? Try 400MHz instead to see if that bumps up your score.

 

I could take my rtx 2080 vram to +1500MHz on OC stable, ask @IllusiveMan that was some crazy good vram hp used. And no it made little to no difference in power draw, but made a 10-15% difference in gpu memory related tasks...

My core was utter trash only allowing an overclock of +150MHz any higher and it would crash. I didn’t win the silicon lottery this time that’s for sure

 

The 180W Vbios will work perfectly. I would still highly recommend replacing the thermal paste with something better before tdp modding...

 

I accidentally flashed while having an external monitor connected. This killed my Displayport

Also, the 180W bios in the thread is an .exe that says "no support on this plattform"

 

No, DisplayPort is the only thing that doesn’t function on the modded vbios. You didn’t kill anything don’t worry.

That’s the problem, if you can source a 180w vbios rom file then flash it. Otherwise your stuck with the 190w or 200w options.

 

Phew, I was really worried. Unfortunately then the 190W bios is not for me. I need that Displayport for my Rift S.

 

I've ran fur mark for a solid 10 minutes and it's completely stable at +150/+150 for overclocks at 69C. CPU hovers around 80-85 under load, sometimes 90. Might spike to 94 but Throttlestop doesn't show any thermal throttling on either the CPU or GPU.

Just for reference.. Yes that's almost as good as a 2080 SUPER




**Going to revise my statement. While the kryonaut stuff is very good, and the benchmarks I've ran were great, gaming was very different. Modern Warfare pretty much caused everything to skyrocket. Not sure why that was so demanding but it definitely throttled then blue screened so it looks like LM may be the only way to get it under control.

 

Flashed my stock bios with a HDMI monitor connected. Dispayport was working again. Flahes the 190W. Diusplayport was dead again. Flashed the 150W bios. Displayport is still dead!!

I checked the cable and the monitor with a diffrent laptop and there it is working fine. So the connector on my Omen 17 must be broken

edit: I had a Thunderbolt Dock commected (HP G2) the last time
i flashed. Its dp commections are also no longer recognized.

edit2: OK, looks like that's a driver problem. After flashing a new bios I have to do a clean uninstall with DDU and install al NVIDIA drivers again. Then both Displayports are working on the stock bios.

 

Does liquid metal does a permenant dmg on the gpu ? Coz I spilled just a drop and the laptop was turned on but then I cleaned it with alcohol and repasted with kryonaut is there a permenant dmg on the gpu now ? Should I sell it from now ? Coz now it works fine I dont know what to do.

 

If everything is working fine and you did clean up properly there should not be any permanent damage.

How did you remove the cooler? I loosened the two screws on each of the fans and the four screws on both the cpu and the gpu cooler. But the cooling unit was still stuck and did not move at all, even when applying gentle force. .

 

If you suspect you may have damaged your laptop, then selling it would be dishonest and unfair to the buyer unless you were clear upfront that you spilled LM onto the PCB.

 

You'll need to apply more than some gentle force. Actually, if you can get a small plastic spudger or pry bar in between the heatsink and PCB, that should help you lift it up and off the CPU/GPU dies. It's actually somewhat similar when getting the back cover off but not near as much force.

 

Start with the part of gpu heat spreader closest to battery, if that doesn’t work, prying there and the part of cpu heat spreader closest to the memory at the same time should do.

 

Im definitely not selling it , its works better than other laptop I have ever seen the weird thing is that im able to undervolt cpu core and cache to -.250 which is I believe unheard of and thats at 38x speed for all cores but the weird thing is to make it stable for that undervolt it has to be plugged in , otherwise I have to use -.150 undervolt which is really weird I dont get how does making the laptop plugged in helps stablizing the undervolt ! And im really happy with the laptop I havent seen it reach 70c yet oh and gpu still didnt reach 60c.

 

I am afraid of damaging the laptop when doing the repaste.

However, I don't understand why the CPU is overheating this much. GPU is fine and stays within the low 70ies even with the 190w bios.

In High Performance mode the CPU will always reach 90-100°C and sometimes starts to throttle. This is with 16-17°C ambient temperature!
Balanced mode is fine with CPU temp in the 80ies and Core Clocks of around 3.

 

OK, I am going to try LM again. Are there any other screws to be removed then the ones I marked? Do I have to remove the fan covers?



Edit: I used a squeegee to pry the heatsink loose. Boy, did they spill alot of thermal paste around. With Liquid Metal temps are down from 96°C in 3DMark FireStrike Physics to 74°C. No kidding.

 

Just be careful with the LM and that you have it masked off well. I tried using 2 different types of LM and both didn't do anything for me as far as temps go. I made sure to apply it to the dies and to the heatsink so it would make good contact but the temps weren't anything impressive. Also, since they're copper heatsinks, there's a good chance the LM will get soaked up by the copper during the alloy process the gallium likes to start and you will need to apply more later on. So if your temps start creeping back up, that's the reason why.

Actually the biggest drop in temps were to use the TG Kryonaut stuff with a downclock to the turbo ratio from 48 to 40. Now everything is extremely stable even with the 190w vbios.

 

Yes, I masked them off with insulation tape und applied LM to both the heatsink and the die. It is working wonders for me.

 

LM works great on just about any laptop. Just make sure the heatsink contacts the die directly, and there is no gap. You guys are outperforming my GTX1080 MXM 200 watt GPU. Power modding will do nothing for me. Most I can manage is 1,911Mhz for several hours in a game and this is only consuming about 140-150 watts of power. I am limited by temperatures. GTX10 series runs alot hotter, then RTX20 series.

 

This is great stuff and gives me something to look forward to, thanks DaMafiaGamer. Right now I've had the thing a month and I'm more worried about getting general heat under control. I put two ADATA XPG SX8100 2TB SSDs in this thing and they were smoking until I finagled the bracket over both drives with the heat spreaders on, I had to bend it a little to make contact. I'm still having low bench write issues with them.
My biggest issue now is how to get this CPU under control. It will take a -0.150 mV undervolt and be stable, but it crashes while driving my Oculus. Out of the box with no XTU work, the spiking to 100 C is just ridiculous. Which LM did you use, and do you mind discussing the CPU repaste and your temp results now? I see you are getting about 95% in userbench, mine is about 84% in full meltdown mode. Undervolting improves the temperature but not the score on this, so the factory paste job seems fine for the GPU (never over 75) but grossly inadequate to the CPU.

 

Sounds like you need some thermal grizzly conductonaut liquid metal for the CPU. Tape around the CPU die with “Kapton electrical tape” for protection. Then apply some LM to your cpu heatsink area, rub the LM on the cpu heatsink area until it is soaked in to the heatsink really good. You are Practically painting the LM in to the heatsink. Then apply a thin layer directly on your cpu die.

I would run LM on the GPU too, and follow the exact same steps as above. The lower the temps, the less power it will consume. Meaning more frequency and more boost inside of this 190TDP envelope.

You can get by on the GPU with standard Kryonaut thermal gel, as RTX series runs pretty cool already.

But, lowering your GPU temps as much as possible will only help your total power consumption and lower the entire systems heat.

You could also go for Fujipoly thermal pads on the entire laptop. This will drop overall temps even further. Replacing thermal pads is a tedious process, and higher quality thermal pads like Fujipoly are extremely dense and much harder, so they do not squish down very easily like the cheaper factory oem thermal pads. So running the precise thickness is paramount for optimal temperature!

 

I'm a bit intimidated by the LM, I've read it's material sensitive and I assume this whole HP set is solid aluminum. Anyone notice if the contact patch is copper (like my old MSI)? I have plenty of pad leftover from my last repaste project, lemme lookup on Amazon which thickness was needed for the MSI GT.
I'll probably just try with MX4 first to see if that does the job. I'm more interested in having the thing perform the way it's supposed to out of the box without screaming, rather than squeezing 110W continuous out of the CPU. (Someone actually managed that according to the XTU benches).

 

The cpu area is probably copper being a high end laptop. LM is the way to go! You’ll be fine.

 

Yeah, it's all copper.

The thing with LM is you REALLY can't get it on anything else besides the CPU/GPU dies and PCB. If you decide to go this route don't forget to apply some kind of acrylic conformal coating around the GPU die's resistors and capacitors. You will also want to mask it off with kapton tape. Same with CPU. This actually saved me from the pressure of the heatsink mating with the dies causing the LM to spill out from the corner and sides. There's tons of how-to guides on how to apply the stuff and I highly recommend looking those up before doing it. Also, it WILL stain and may pit the copper of the heatsink. You won't be able to get it off since the gallium is forming an alloy with the copper and it will set inside of the copper. This wont affect the heat transfer capabilities but it will turn colors from a copper orange to silver.

LM is easy enough to clean up with rubbing alcohol should you make any mistakes.

I've applied this stuff 3 times now to my laptop and I just cannot make it work. I constantly see it spiking at 100C on the CPU. I'm beginning to wonder if the heatsink isn't making good contact pressure on the CPU? TG's Kryonaut stuff seems to work really well but maybe that's because it's a thicker paste?

 

Just wanted to chime in and share my thoughts and experiences. I've recently acquired my Omen 17 with a 9750H and RTX 2080. I didn't manage to see my scores and temps with stock paste but I did get it repasted with LM and K5 Pro. Temps are 70 on GPU and never reaching 80 on the CPU with the 200W bios. Firestrike score is 20350 and Time Spy is 9350. Does that sound right? In addition, I noticed there's a ~15 C difference between my cores. Is that normal? Although the core temps are great (one core being ~64 and the hottest being 79) I feel there is some concern. Otherwise, great machine for ~$1350.

 

Thanks for your replies--I adjusted the screws a bit when I was messing with the SSDs last time, and after the BIOS update (supposedly security only) and most recent Windows update it seems to run a little more temperature stable, sawtoothing between 40 and 80 (this is in comfort mode, a 45W cap). I agree with the core spikes, as I have 8 backblaze threads running the activity jumps among all the cores and at any given moment one is much hotter. The general temperature reported seems to be a bit melodramatic.
I've run in to issues with the undervolt and have given up. It's crashed a few times, one causing the 2080 to not be recognized which I salvaged with system restore. Since -0.05V caused Oculus to crash I think I may just have a lame chip. I'll give it some continued use with no undervolt to make sure it's generally stable before revisiting.
Thanks for the LM advice, I have next week off and my try repaste #1 with what I have on hand before deciding if cryonaut is the way to go. MX4 did well enough on my 4810MQ which had the same issues of these awful temperature spikes, and the MSI GT 2014 models had a pretty weak CPU cooler.
You nailed it on the head, for the price I knew this thing would be a project but I saved probably $2000 over what I really wanted with identical components down to the 4K panel.
I have a whole list of other questions about this machine but maybe they belong in a new thread: Anyone have 2 nvme drives that bench properly? My write speed is halved, and I'm not sure if it's because they're hot or if they're not both running in x4 (HWinfo says they are). Anyone installing most current nVidia drivers? Windows did it for me and I think it may have contributed to the need to restore, so it's back on HP's July version.

 

It was actually confirmed by a few people in another thread where the 441 driver caused issues when undervolting with Throttlestop. I too had the same issue where the system would crash and the GPU would be undetectable. Only way to fix it was using the HP Recovery Tool. After installing 441.87, it appears this issue has been resolved.

 

Thanks for that! Great tip, the performance is so good as is I've disabled Windows auto driver updates so down the road I'll just use the HP ones as they filter down, or as recommended on here.

 

No, I'm not sure about Firestrike but Time Spy is about 1000 too low if you're running a 200w vbios. Temps sound about right but I think you need to adjust your ratios or possible power limits. Check throttlestop to see what might be wrong or to adjust values. In Timespy you should be seeing 10k+.

If you have any questions just post or shoot me a PM and I'll try to answer back base on my own experience. Not sure about the nvme drives. This laptop seems to be EXTREMELY picky about what m.2 drives you put in it. It's like they MUST be nvme drives and not a simple M.2 2280 drive like the WD blue. I'm using the latest nvidia drivers and they seem fine.

There's actually an easier way to fix this. You don't need to use the HP recovery tool BS. In Device Manager, look for the PCIE display device and uninstall it. Then try searching for any new hardware if windows hasn't picked it back up and it will install it as a 2080. This issue isn't exclusive to the new drivers, I've had it happen a few times regardless of what I installed.

 

Yeah I thought so. Will follow up and try again. As mentioned, the 441.87 WHQL drivers seemed to have made the problem go away. The previous driver was said to have been the culprit as research led me to believe. I think your issue was completely different. I couldn't even boot into OS nor Safe Mode and it forced you to reinstall.

 

What is the graphics score? I assume that score is for the full benchmark? And they sounds very good if you haven’t tweak it.

 

Oh okay, yeah I never have had an issue trying to boot into the OS but windows has "lost" the video card for whatever reason. Was freaking out the first time it happened.

 

I just wanted to post thanks! I am using the 190W vbios and it works like a champ!

 

I did see lower scores with the 200W then with the 190w bios. I don't know why.

Try the 190W bios, set energy plan in both Windows and Omen Contorl Center to High Performance and bench again. also make sure VSYNc is disabled.

 

Just an FYI -

I am using the 190W vbios and like some have mentioned, the mDP (displayport) does not work. However I use a Dell Thunderbolt Dock and it works great. My external display is an 240Hz Gsync display that works great. Gsync works on it too.

 

Does ist work at 240Hz over USB-C? I am using the HP dock and it only supports 120Hz.

 

It's actually a full thunderbolt 3 dock, not USB. I am guessing that is the difference.

The model is Dell TB16. You can get them relatively cheap used and I have plenty of extra dell AC adapters in case you buy a cheaper one without it. It also worked great on the Acer Predator Triton 500 I had.

 

Thought I'd give a quick update. Thanks to everyone's advice I managed to repaste with MX4 and it's running much cooler and quieter.
The rest of the Omen17t saga is more problematic: The boot SSD just died, totally unrecognizable even in an enclosure to my ultrabook. I'm sending it back for a refund and will just get a cooler QLC SSD since I think the Omen case doesn't have good airflow through it except straight to the fans and out. Thankfully I had a fresh Reflect image which restored easily to the other one which is now the boot.
I've downloaded the vbios mod and will save it for the warranty expiration, unless I get desperate for GPU performance. That seems to be the only thing on this that _is_ cooled adequately. Given that the baseline for the CPU cooling is the GPU's idle TDP, thanks to the heatpipe design, I'm reticent to make the right half of it any warmer. It's a brilliant hack, though.

 

That's a steal, I added the 4K and color keyboard back on Black Friday and it was about $1800. That Intel H10 is trash though, be careful with it. It lasted about a week and started getting corrupted, wound up using my add-on as the boot and rebuilding from recovery drive, a lot less HP junk.

 

Not entirely sure how to upload pictures here, but here are my Firestrike results:

200W: 19788
Graphics score: 24132
Physics score: 17641
Combined score: 9131

190W: 19678
Graphics score: 24229
Physics score: 17648
Combined score: 8800

This is with the balanced power option and performance profile in the Omen Command Center. Everything else is stock.

 

200W: 20332
Graphics score: 25436
Physics score: 17657
Combined score: 8927

With performance mode on both battery plan and command center.

 

Nobody on the ailienware forum would dump the 180w rom for me so I decided to see if I could get it myself. The executable wont run on our machines but I was able to extract it and look through the data files. I did a hex compare against the 190w rom on one of the files big enough to fit a rom file and found the 180w rom embedded in plain text. Matched it up byte for byte using the headers and voila.

Just flashed it on my machine and its working fine, still no display port though which is the reason I wanted to try it. I haven't done any benchmarks because I never took any with the other bios's to compare with but would be interesting to see others results.

 

If the 200W vbios is generating more heat, I don't think it's worth it. I'm running the 190W and got these results:

Total: 21490
Graphics: 26823
Physics: 21212
Combined: 8696

Can get a SS if needed but I think if you OC and undervolt it a bit more you can get higher scores. Not sure if it really matters or translates into actual games but it's still nice.

 

Ran timespy with the 180w bios.
Stock graphics score: 10104
OC (+175/+500) graphics score: 10879

Core clock hovers around 1900MHz with the OC, max temp 70C with stock thermal paste.

 

Firestrike on 180w bios. Had to drop the core to +150 otherwise it would crash.

Graphics: 27183
Physics: 19524
Combined: 9261

Core hovers around 2000Mhz. looks like dx11 boosts higher for some reason.

 

Sounds good. Haven't gotten around to tweaking it much yet. I mentioned the temps earlier so I'm not really concerned about the heat with the 200W vbios. You have the i7 or the i9?

 

The i9

 

I wanted to ask you about this, or anyone else who may know: Do you think this would benefit from a shim on the CPU? What thickness would you recommend? The benefits of my repaste have diminished over the past month, so I think MX4 may be too thin for the mounting pressure, plus if it hasn't squeezed out it may be cockeyed due to your mentioned alignment issue between GPU and CPU.
Thanks!

 

wow nice find... just for info, this can be done for every rtx laptop? my rog 704gw with 115w 2070 can be pretty happy with an 135w or 150w vbios for example, just to hit the 1900mhz as stock desktop cards...

 

i did the 190w and 5 cores from 9750h are at 100% usage all the time the other 7 are running 13-51% whats going on?

UPDATE: Figured it out

New issue max turbo is 4.3ghz all core. doesnt go above that. i think its my cstates