1610ftw

There is a bunch of changes that you have to make in bios to enable undervolting / overclocking, have you done that? If not please check out the videos below. The second one goes through all settings I believe and the other one os more geared towards the GT77 but I am not sure if it gives step by step instructions. Personally I prefer to do everything in Throttlestop instead of the bios or Intel XTU. It also allows me much easier tweaking and as my laptops usually only get rebooted once a week if at all I do not mind having to start Throttlestop manually and there are also easily switchable presets and even an automatic switch between battery and AC power if you want to. In any case it is a very handy tool that also gives me nice info in the task bar about power usage and CPU and GPU temps so it is what I always use. I just hope that things are not so borked that you will not be able to use XTU / Throttlestop as that is horrible when on a power budget!

1080SLI performance: Easily can do between 15 and 16000 points TimeSpy, without any shunt-modding or other excessive measures needed, just get your cooling straight and use MSI afterburner. Finally surpassed by a single card 3 generations, 5 iterations and more than 6 years later, that is a very long time! 2016: 1080SLI 2080 2080s 3080 3080ti 2023: 4080/4090

Always nice to be mocked by level 1 support after the machine was messed up through their firmware updates. From what I have seen MSI used to be pretty good with allowing CPUs to attain the maximum power the system could sustain and maybe I could understand them setting the long term limit to 160 or 150W with these new CPUs that in theory could be set to consume south of 200W which is just a bit too much but going down to 120W is a waste of the 13980HX. For now you may want to check how much you can undervolt your GT77 with the new target of 120W - it should still give you quite good performance even with 120W and an optimized undervolt should at least give you 10% higher performance than stock so if you did not undervolt before your performance will probably be quite similar and not suffer that much. I would still try to roll back those recent "improvements" but for now that should help.

How would that even be possible when you are using Throttlestop and an unlocked 13900HX?

I never game so no gaming benchmarks 🙂 My stock 2080 with 155W is 10.2K. Stock 2080 with 200W would be closer to 11K.

Thanks a lot, a lot of work went into this and it is nice to see it documented!

Nah, that would be too easy 😄

Yes indeed, for a non-gamer like me who already lost G-Sync with a new panel it would be quite tempting 😄

I doubt that dynamic boost will be much of an issue, probably a 5 to 7% reduction in performance but still better than the 2080 Super. The lack of GSync could of course be an issue for people who need it but everybody has to answer that for himself.

Obviously it would be a modification but then the 3070 and 3080 are running on P775, P870 and X170SM-G chassis of several members so it is possible. The question is if you need it - if temps are fine and you are at mostly gaming with FHD resolution I would not expect it to be that big of an upgrade but if you want maximum power savings / a quieter system or could use some added GPU memory then it may be worth it.

I think best was something like 11365 but that was the best I could get after many tries and with afterburner activated. Edit: I could not find the old scores but here is my best result today with the P775 with MSI Afterburner and at 155W, about 7% more than without: So it is less than a 400 points difference compared to the MSI or about 3.5%, not that much really. No delidding here - I would first want to find a CPU that is worth it and in need of a delid. Probably less gained if you already have a sub 10 degree temp difference between cores and much more if one or more cores get a lot hotter than the rest. In my experience it is OK to run a bit higher in Time Spy than in real life so I regularly use at least 4.8 for Time Spy and I think it could probably be something like that for games as games are usually less demanding on the CPU than a multicore benchmark run. You got to keep in mind that 9900K are quite wide ranging with regard to power consumption and therefore temps and performance. My best 9900K is at least 30% less in power consumption at 4.6 GHz than the worst one that I have seen - would not want that bad one in a laptop! So the short of it is that these can get VERY hot and there are three things to look for: - binning - cooling - undervolting That combo can easily make a difference between 4.4 and 4.8 up to 5.0 GHz whereas I am not so sure how much better you can get with delidding when you have an already very good CPU. As for the 8700K it is probably pretty good for your use case and it does not need that much power compared to a 9900K but 4.5 seems a bit low and I would expect more after you have done some optimizations. How much higher depends on all the above factors and games are a rather tricky load where both CPU and GPU are working hard and I usually only deal with that in benching.

OK, did a quick run without MSI Afterburner and I got 10.2K with 155W max which is the same power consumption as in my MSI GT75 that tops out at 11300+ HWinfo data: The 9900K in the P775 is not exactly great so I am currently thinking of putting in another one with better binning but for this benchmark it made no difference as I limited it to 4.0 GHz all cores. For comparison on a well set up system of that generation (Coffee Lake) I would try to run the CPU at 4.8 to 5.0 Ghz all cores.

Can't remember what it did but can have a look without MSI Afterburner activated, probably a bit above 10K. I think it is also the 150W bios and I would have to do something about that to get really good scores but cannot be bothered at the moment - too many laptops too little time 🙂

I agree you should do better than that. My P775 went to about 11000 with a few small adjustments and I left it at that for the time being as I am more of a CPU guy but you should see that, too. So most probably the heat sink not making good contact and I would also check the pads and if you are not certain go with some putty from Upsiren or something like it where the exact thickness does not matter,

Yes it is, 150W 3080 > 200W 2080 Super. The difference may not be big in raster performance but as per the post above mine there are other advantages and when one has to order a GPU anyway and the 3080 is not much more, why not? I also just checked and if you look at the 12509 TimeSpy score from @win32asmguy you can see it is within 0.06% of number 2 in the all time 3Dmark rankings for the 2080 Super and only @Prema in number 1 position is really better and I am pretty sure that is without making much of an effort: Best I got out of my 2080 Super in the SM-G was a little over 12200 and I would expect to get to at least 13500 on the 3080.

you may want to look out for one of the unified heatsinks for it that Eurocom sometimes has up for auction: https://www.ebay.com/itm/116123553696?mkcid=16&mkevt=1&mkrid=711-127632-2357-0&ssspo=_LxgfDnLRKa&sssrc=2047675&ssuid=&widget_ver=artemis&media=COPY

I think this was a barebone and it did not have a GPU so why not? It will give better performance than the 2080 Super with less heat and double the memory so a win-win-win.

Pretty sure that you will want PTM7950 for the 10900K 😄

Not sure how much power that 2080 is getting but I would usually get between 11200 and 11300 out of the 2080 with a power draw of up to 155W and others have gotten up to 12000 at 200W. So performance is quite a bit higher for the 2080 and I would say that watt for watt it should give an increase in performance of at least 40% over the 1080 with good results even with low power consumption which is nice for temps and longevity. Here is the leaderboard with 9900K and 2080: https://www.3dmark.com/search#advanced?test=spy P&cpuId=2402&gpuId=1244&gpuCount=1&gpuType=ALL&deviceType=ALL&storageModel=ALL&showRamDisks=false&memoryChannels=0&country=&scoreType=graphicsScore&hofMode=true&showInvalidResults=false&freeParams=&minGpuCoreClock=&maxGpuCoreClock=&minGpuMemClock=&maxGpuMemClock=&minCpuClock=&maxCpuClock= I think I got 11700, too with my Alienware but was a bit disappointed at the time that I could not get it up to at least 11800 so my results are not in there 😄

I think you're set, just be careful to not overdo it with the memory timings and check that you get everything running first. If you like it clean you could go for some kind of carbon / graphene pad for heatsink testing and positioning and later replace it with PTM7950 but any thermal paste will do if you don't mind a bit of cleanup before applying PTM7950.

@win32asmguy has a 3080 running in his SM-G. I think it might just work when you have the right heatsink and if not it should not be that much of a stretch to get it going. Gsync may be an issue but then even that should work with the right GPU / display combination.

Great news and to my knowledge the first time that somebody has this display running in a P775 or P870 on this forum. You mention Freesync, do you have an AMD GPU running or was that a typo and you meant GSync? In any case it would be great to know for others which cable you used and which GPU and of course the vbios.

PTM7950 FTW!

That is a really nice display upgrade! Not sure if there is reason to believe that the display is defective but in case it isn't have you checked if you can see the bios via an external monitor? After all you can potentially get a display output with HDMI, mini DP and USB-C/TB3. If one of those options was working that would be good enough for me and I would keep the display as it has QHD resolution and it is superior to the old QHD option that only was a TN panel with a lower color fidelity.

Nice - going up again! If you already have the PTM7950 don't sweat it. I have both times gotten mine from non-official sources and THINK it was legit but cannot be sure - the results at least seem solid and repeatable and yours look good, too!