Dell Precision 7670 & Dell Precision 7770 owner's thread

[RockoDesvan]

On 9/10/2022 at 11:41 AM, Aaron44126 said:

I've been trying to figure out why I get some inconsistent framerates and audio stuttering from time to time with Final Fantasy XV (now about 25 hours in).  I noticed some pretty significant DPC latency, coming from the NVIDIA driver nvlddmkm.sys.  The latency incurred is at least 5ms — I stopped running LatencyMon after I saw a huge spikes, I didn't wait to see if any higher ones popped up.

 

I got it to drop down to <1ms but only by forcing the NVIDIA GPU power state to P0 using NVIDIA Inspector.  So, the latency is likely caused by NVIDIA GPU power state changes (something that I have fought with before).  Note — Setting "Prefer maximum performance" in the game's profile in NVIDIA Control Panel was not enough to stop the issue.

 

Something to keep in mind if running real-time audio or other time-sensitive applications while the NVIDIA GPU is active.

 

I now have tooling built into Dell Fan Management which, when combined with Process Lasso, can switch the thermal profile (i.e. from "Quiet" to "Ultra Performance") when certain applications are launched.  Now I need to do something similar for the NVIDIA power state...  [Edit] Got it.

That awesome! I checked your github but the latest release of you app dates back to April of this year.
Do you have plans to make a new release of your Dell Fan Management app with this new capabilities? 😄

[Aaron44126]

5 minutes ago, RockoDesvan said:

That awesome! I checked your github but the latest release of you app dates back to April of this year.
Do you have plans to make a new release of your Dell Fan Management app with this new capabilities? 😄

 

I've dropped some beta builds in the thread (linked in signature)...  I really should get everything together and do a proper release, just another thing on the "can't find the time" pile.  😕

[-batab-]

Has anyone of you benchmarked (or sort of) the igpu of the HX series? 

 

How wonder how much slower it is when compared to the iris of the H series? 

[1610ftw]

On 9/2/2022 at 9:52 PM, TwistedAndy said:

 

Actually, you can undervolt the CPU with Hyper-V active. If you're planning to use ThrottleStop, you would need to disable VBS (Virtualization-Based Security). This feature locks some registers and prevents them from changing.

 

Once you find the stable voltage offsets, you can put them directly to the BIOS settings in the same way as you unlock the undervolting. In this case, you will be able to enable VBS.

 

On my device (XPS 9720) I was able to achieve -0.15V offset, but for some reason the system crashed while my IDE was indexing the project. Maybe, there're some JVM-specific issues there. I ended up with -0.12V offset. As for the system agent, it is unstable at low voltages, so I decided to apply -0.06V for that.

 

It is quite stunning that despite the HX processors being out for a while now there is not one single report of somebody undervolting an HX CPU.

 

Hopefully @Aaron44126or @win32asmguywill get around to try it at some point.

[Aaron44126]

1 hour ago, 1610ftw said:

Hopefully @Aaron44126or @win32asmguywill get around to try it at some point.

@win32asmguy has done it, but P cores only (see page 5).

[-batab-]

3 hours ago, 1610ftw said:

 

It is quite stunning that despite the HX processors being out for a while now there is not one single report of somebody undervolting an HX CPU.

 

Hopefully @Aaron44126or @win32asmguywill get around to try it at some point.

I've found some people that did it while lurking around over reddit/youtube/throttlestop threads here and there. Not many though. However, I can't really remember exactly where I found stuff so I can't pinpoint everyone I stumbled upon in the last couple of weeks.

 

One guy was certainly talking about how undervolting the legion i7 (lenovo legion subreddit) boosted his gaming performance consistency by (especially) lowering his stuttering.

 

If I'll pull the trigger on a legion i7 in the next few days I'll certainly do that.

[1610ftw]

2 hours ago, Aaron44126 said:

@win32asmguy has done it, but P cores only (see page 5).

Yeah, forgot to mention all-core undervolt but that probably opens another can of worms.

In any case almost 15k CB R23 for the P-cores with only 75Watt is a huge jump compared to previous CPU generations.

[1610ftw]

53 minutes ago, -batab- said:

I've found some people that did it while lurking around over reddit/youtube/throttlestop threads here and there. Not many though. However, I can't really remember exactly where I found stuff so I can't pinpoint everyone I stumbled upon in the last couple of weeks.

 

One guy was certainly talking about how undervolting the legion i7 (lenovo legion subreddit) boosted his gaming performance consistency by (especially) lowering his stuttering.

 

If I'll pull the trigger on a legion i7 in the next few days I'll certainly do that.

 

With E-cores not going anywhere soon it would be cool if Throttlestop could introduce separate settings for P- and E-cores.

[-batab-]

1 hour ago, 1610ftw said:

 

With E-cores not going anywhere soon it would be cool if Throttlestop could introduce separate settings for P- and E-cores.

 

Unclewebb is still quite active on the TPU forum so we might get an answer over there. Considering how many things he added since I got my 7820hk, I'm faithful. 

 

To be fair XTU has also gotten better. It went from utter useless to an almost usable alternative. 

 

EDIT: damn, those 13900K multithread scores almost make me want to wait another year 😕

[1610ftw]

1 hour ago, -batab- said:

 

Unclewebb is still quite active on the TPU forum so we might get an answer over there. Considering how many things he added since I got my 7820hk, I'm faithful. 

 

To be fair XTU has also gotten better. It went from utter useless to an almost usable alternative. 

 

EDIT: damn, those 13900K multithread scores almost make me want to wait another year 😕

 

Tried XTU a few times but I found Throttlestop to be superior so hopefully Unclewebb will get around to it.

 

 

 

[-batab-]

6 hours ago, 1610ftw said:

 

Tried XTU a few times but I found Throttlestop to be superior so hopefully Unclewebb will get around to it.

 

 

 

 

I 100% agree. And considering how much more you can get out of top notch cpus with such apps I really hope the support won't ever stop. I would even be willing to pay decent money since it's no different than paying for a better piece of hardware for me. 

 

In fact, I actually wonder why no producer goes a step ahead and introduces an automatic routine to produce an optimized Throttlestop/XTU/bios profile for some bucks. This is usually worth more than a cpu upgrade. Maybe it's just my nerd attitude but I really wonder why. 

[1610ftw]

11 minutes ago, -batab- said:

 

I 100% agree. And considering how much more you can get out of top notch cpus with such apps I really hope the support won't ever stop. I would even be willing to pay decent money since it's no different than paying for a better piece of hardware for me. 

 

In fact, I actually wonder why no producer goes a step ahead and introduces an automatic routine to produce an optimized Throttlestop/XTU/bios profile for some bucks. This is usually worth more than a cpu upgrade. Maybe it's just my nerd attitude but I really wonder why. 

 

Much better to count on most people being sheep and being happy with whatever they are given - that way a CPU that could pull 100W or more often is reduced to 45 or 55W these days which guarantees fewer returns not only on the CPU itself but also on the motherboard it is soldered, too. Plus fans can stay silent at almost all times.

 

Just look at the hoops that our 7770 owners have to jump through this generation which seems to be due to both electronically limiting power delivery to the CPU and also due to it thermal throttling much earlier. It is also easier to hide a bad paste job when the CPU is limited to a lower power delivery.  

 

That being said @win32asmguyhas done all future owners a great service by digging deeper into the CPU side of the 7770 chassis so that CB R23 results have increased by about one third which is fantastic.

[-batab-]

42 minutes ago, 1610ftw said:

 

Much better to count on most people being sheep and being happy with whatever they are given - that way a CPU that could pull 100W or more often is reduced to 45 or 55W these days which guarantees fewer returns not only on the CPU itself but also on the motherboard it is soldered, too. Plus fans can stay silent at almost all times.

 

Just look at the hoops that our 7770 owners have to jump through this generation which seems to be due to both electronically limiting power delivery to the CPU and also due to it thermal throttling much earlier. It is also easier to hide a bad paste job when the CPU is limited to a lower power delivery.  

 

That being said @win32asmguyhas done all future owners a great service by digging deeper into the CPU side of the 7770 chassis so that CB R23 results have increased by about one third which is fantastic.

 

Yeah, on a workstation-class laptop this is quite ridiculous. As if 100% cpu load scenarios wouldn't be possible or even frequent.

 

Not even commenting on the GPU side of things where a 3080ti is almost equivalent to a 3070ti due to power throttling. The real difference is usually in the 300-600€ you get charged for it. 

 

 

[serpro69]

Part of the problem the way I see it is - there's very little competition in workstation-class laptops, and it's getting thinner by the year.

So long as they (OEMs) don't need to compete very hard for buyers, they will keep pushing all sorts of crap on us, because, well, they can and we don't have a choice. So why wouldn't they sell you a more expensive GPU, even if it doesn't do you any good.

To be honest I'm kind of disappointed with Dell this year. Not that I wouldn't get me a 7770 (provided my sales rep ever comes back from vacation and responds...) Then again, at this point I also don't see any good alternatives. But looking at the issues with this year's precisions, starting from manufacturing issues and ending with the hoops that users have to jump through to get the performance up to what it should be when these units come out of the factory... And this is for the top-of-the line business workstations? So much as I like(d) Dell in the past, I'd probably jump ship and get something else ...if there was anything else worthy of getting. 

 

[win32asmguy]

8 hours ago, 1610ftw said:

Much better to count on most people being sheep and being happy with whatever they are given - that way a CPU that could pull 100W or more often is reduced to 45 or 55W these days which guarantees fewer returns not only on the CPU itself but also on the motherboard it is soldered, too. Plus fans can stay silent at almost all times.

 

Just look at the hoops that our 7770 owners have to jump through this generation which seems to be due to both electronically limiting power delivery to the CPU and also due to it thermal throttling much earlier. It is also easier to hide a bad paste job when the CPU is limited to a lower power delivery.  

 

That being said @win32asmguyhas done all future owners a great service by digging deeper into the CPU side of the 7770 chassis so that CB R23 results have increased by about one third which is fantastic.

 

This machine is definitely more challenging to tune. Over the course of a week the temperature deltas had grown by another 10C (71C coldest, 100C warmest) so I pulled the heatsink back off and it appears the paste had pumped out. Repasted and reassembled and its back to where it was before. I am also ordering some Honeywell PTM7950 from China which is supposed to be more resistant to pump out and offer better thermal transfer.

[-batab-]

3 hours ago, win32asmguy said:

 

This machine is definitely more challenging to tune. Over the course of a week the temperature deltas had grown by another 10C (71C coldest, 100C warmest) so I pulled the heatsink back off and it appears the paste had pumped out. Repasted and reassembled and its back to where it was before. I am also ordering some Honeywell PTM7950 from China which is supposed to be more resistant to pump out and offer better thermal transfer.

 

i've had Kryonaut and TFX pump out but solved with SYY 157. Kryonaut actually pumped after long time (although laptop was brand new), TFX after a few days (but laptop was old already). SYY is still working after months and was applied when the laptop was old already.

 

(referring to my experience on my current clevo p650hp6 if it matters)

[TwistedAndy]

On 9/15/2022 at 10:40 PM, -batab- said:

If that's the typical use case why not simply opt for a 6900hx which is way more efficient than Alder lake when below 30W and has close performance in the 40-70W range. 

 

If someone plans to not ever exceed the 70-80W for cpu only loads you basically get better thermals, noise and battery life at a lower price if you just pick up an amd laptop. 

 

AMD 6900HX can compete with Alder Lake H/HX up to 45W sustained power limit.

 

Alder Lake shows significantly better results on higher power levels because of better performance scaling. Actually, it can scale the performance effectively till 80W-90W. In case of AMD 6900HX there's no reason to go above 50W.

 

For instance, on 70-80W sustained power limit, 12700H/12900HK can get ~18000-18500 while the AMD can get ~14000-14500 CB R23 MT. In other tasks the situation is nearly the same. Alder Lake is 20-30% faster.

 

In the case of HX series, they start showing some performance advantage over regular H series on 80+W power limits

 

So Alder Lake gives you more options here. You can have decent thermals, noise levels (<40 dBA) and slightly better performance than AMD at 45W. Or you can set the power limit to 75W, get 15-25% performance boost, and higher noise levels.

 

As for the battery life, it is heavily depends on the particular laptop model and optimizations. Usually, laptop vendors doesn't care too much about that, but Dell did a good job in XPS 15 and 17. I get nearly 10-14 hours on my XPS 17 on light use.

[1610ftw]

17 hours ago, win32asmguy said:

 

This machine is definitely more challenging to tune. Over the course of a week the temperature deltas had grown by another 10C (71C coldest, 100C warmest) so I pulled the heatsink back off and it appears the paste had pumped out. Repasted and reassembled and its back to where it was before. I am also ordering some Honeywell PTM7950 from China which is supposed to be more resistant to pump out and offer better thermal transfer.

 

Have you ordered the pad version or the regular paste? Looks like Lenovo is using it with some success and Lenovo users even went back to it after trying other options that supposedly were better but not as long-lasting.

 

In any case I am not sure how much Dell is to blame with the new heatsink design and how much the new chips are responsible. Looking at Lenovo and MSI where it is seemingly possible to sustain about 120W or more to the 12800/12900/12950HX chips it looks like it may be the new heatsink design and not the chip itself causing the lower performance.

[cm202209]

Hi All,

 

I just got a 7670 with i9-12950HX, A2000 delivered on Friday. When I hook up using WD19DCS doc, dual TB cable and set for hybrid graphics, not discrete external output (meaning the external video seems to be driven by UHD), I see flickering on my external display (U4320Q) when connected to either display port, HDMI or USB-C. I don't see the flickering when I change bios setting to use discrete graphics for external ports.

 

I've updated all graphics drivers, bios is the latest etc. 

 

Any thoughts? Is this a driver issue, bios issue or is this a hardware issue with UHD/Processor?

 

Thanks.

[Aaron44126]

21 minutes ago, cm202209 said:

Any thoughts?

 

Well...  It's a Dell laptop, a Dell dock, and a Dell monitor, so that's good for getting Dell to fix it anyway.

 

Make sure that you have the latest dock firmware installed.  Release notes mention monitor compatibility.

https://www.dell.com/support/home/en-us/drivers/driversdetails?driverid=5x3w3&oscode=wt64a&productcode=dell-wd19dcs-dock

 

Clearly, it is an issue with the display signal coming off of the Intel integrated GPU (since the discrete GPU doesn't have the problem), but I don't know if it is something that could be fixed by doing, say, a motherboard swap... it might be the case that they are just "all this way" and the fix is going to be more on the firmware/software side.  I'd refer you to Dell support to see how they want to handle this, be it swapping out hardware or trying to reproduce the issue in their lab to work on a fix.  (This is the sort of thing that will be more smooth if you have ProSupport.)

 

Have you tried connecting your U4320Q monitor directly to the system via USB-C to see if the flickering shows up that way too?

[cm202209]

Just now, Aaron44126 said:

 

Well...  It's a Dell laptop, a Dell dock, and a Dell monitor, so that's good for getting Dell to fix it anyway.

 

Make sure that you have the latest dock firmware installed.  Release notes mention monitor compatibility.

https://www.dell.com/support/home/en-us/drivers/driversdetails?driverid=5x3w3&oscode=wt64a&productcode=dell-wd19dcs-dock

 

Clearly, it is an issue with the display signal coming off of the Intel integrated GPU (since the discrete GPU doesn't have the problem), but I don't know if it is something that could be fixed by doing, say, a motherboard swap... it might be the case that they are just "all this way" and the fix is going to be more on the firmware/software side.  I'd refer you to Dell support to see how they want to handle this, be it swapping out hardware or trying to reproduce the issue in their lab to work on a fix.

 

Have you tried connecting your U4320Q monitor directly to the system via USB-C to see if the flickering shows up that way too?

Hi,

 

Thank you for the quick response. I don't see any flickering when connecting it directly to either HDMI or the USB-C DP port. Also, in terms of flickering, it is snowy/ghost image splotches intermittently in different areas of the display and then it goes off, almost as if it lost sync and then comes back after 2-5 seconds. I see this when it boots into Windows 11, have not see it when I see the Dell bios splash. Note, I've used this setup with 7540 and 7560 without issues.

 

Just finished running support assist and it says my RTX A2000 GPU failed. Just requested a replacement part.

 

 

[cm202209]

@Aaron44126, thought about it some more, the Supportassist diagnostics never got to Intel UHD and would just stop with RTX A2000, then after running it a couple of more times, got erratic results with random stuff marked as failed, and made no sense.

 

I'm within the return period and so just decided to return it rather than deal with the headache of troubleshooting it. Was somewhat of an impulse buy anyways with some great discounts, but not worth it if it does not work. I have a Fedex label from them, it is on it's way back. Will stick to 7560 for now, that has worked well and continues to work really well.

 

Thanks again for all the help.

 

[mkolejak]

Hi guys, I've been waiting for my 7670 since July 8th, it looks like they shipped it more than a week ago and still nothing. 😄 Does anyone of you have a uhd+ 16 touch screen? Can you tell me what's the exact piece of hardware is that, or / and type of touch screen? If its active, multitouch or whatever, because I can't find any info about that 😄

Thank you for info!

[win32asmguy]

On 9/19/2022 at 4:34 AM, 1610ftw said:

Have you ordered the pad version or the regular paste? Looks like Lenovo is using it with some success and Lenovo users even went back to it after trying other options that supposedly were better but not as long-lasting.

 

In any case I am not sure how much Dell is to blame with the new heatsink design and how much the new chips are responsible. Looking at Lenovo and MSI where it is seemingly possible to sustain about 120W or more to the 12800/12900/12950HX chips it looks like it may be the new heatsink design and not the chip itself causing the lower performance.

I ordered the pad version. Lenovo does seem to have had success with using it. If anything it seems to have a more consistent result from the factory.

 

Also to rule out the heatsink I ordered a replacement from Dell. It took about 30 minutes of troubleshooting then they sent an overnight parts dispatch. The new heatsink has Sunon fans instead of Delta. The specs are slightly different on the fans (Sunon, 5v, 0.8a, 9.4cfm) vs (Delta, 5v 0.5a, 9.2cfm) and they have a different sound cycling on/off which is quieter. The stock TIM also appears to be better than what was shipped originally from the factory, I will use it this week to see how it performs. This one is already exceeding what was sustainable with the original.

 

[Aaron44126]

BIOS update 1.6.2.

- Improved the stability of the AC adapter when connected to the system.

(...Huh?)

 

https://www.dell.com/support/home/en-us/drivers/driversdetails?driverid=7J3NK

[Edit] Updated with no issue.

 

9 hours ago, win32asmguy said:

Also to rule out the heatsink I ordered a replacement from Dell. It took about 30 minutes of troubleshooting then they sent an overnight parts dispatch.

 

Did you repad the new one or are you using the stock pads for now?