Dell Precision 7670 & Dell Precision 7770 owner's thread

[Ionising_Radiation]

7 hours ago, Aaron44126 said:

A dGPU chip which NVIDIA had spec'ed to work at 165W

Frankly, I think part of this is NVIDIA's and Intel's fault—150 W+ power limits were quite mental for a notebook (even a 'gaming'/'workstation' one) back in the Sandy/Ivy Bridge and Haswell eras. I'm not even going back that far; my old Clevo W230SS from 2014 had a 45-60 W limit on the CPU, and a 75 W limit on the GPU (and still took a 120 W power supply). 

 

Intel has (mostly) forgone efficiency and is chasing ever-higher power limits; likewise with NVIDIA. AMD is not even in the notebook scene; I haven't seen a single 5XXX/6XXX/7XXX-series notebook where I live. Lenovo sells some, but only off their Web store. 

 

I would love to see an M1/M2-esque CPU in a non-Mac. But until then, like @Aaron44126 said... We are stuck with compromising. 

 

 

[pintie]

12 hours ago, Aaron44126 said:

Yeah, I think that the main sticking point here is expectations.

 

That... my 7670 ist speced at maximum with A5500 and i9 an it was pretty clear to me that this thing will not work at 300W.

maybe boost for some seconds, but you can't get so much heat out of this chassis.

 

I have to admit - the cooling is even worse then i thought. But this thing is still the best solution for me at the moment.

 

 

@Aaron44126 - yes the happy black and amber Display times... when you could se where your cursor had been 10 seconds ago 🙂

[MyPC8MyBrain]

38 minutes ago, pintie said:

amber Display time

in my early days the only monochrome monitor available was the green kind without mouse support,
pure terminal in the early matrix days😉 

[1610ftw]

13 hours ago, MyPC8MyBrain said:

bit on the fence there, i didn't want to buy a compromise at $4500,
for around 2k i don't have an issue with some compromising,

i was hoping the 7770 larger chassis will have bit more thermal headroom than the 7670 has,
i think i even stated somewhere here that if i can get the 7770 to 13k score it will be good enough for me to keep,
yes i am not that far off from that score but i am well aware that I am off by at least 1400 point from the same hardware average performance and that is a hard pill to swallow coming from Dell top of the line workstation,

 

 

Your best chance to get the highest level of performance you seek from a Dell laptop (instead of going for the other brand choices) is to wait for the new 18" laptop coming out early next year. That should give you an at least 15k+ Time Spy and a 32K+ CR23 within your budget if you choose the 13900HX and the 4070 or more if you go for the 13980HX, 4080 or 4090. Warranty may be limited to 4 years max though.

 

As others have pointed out even if this kind of hardware makes it to the Precision line it is most likely that they will stay with this years hardware in the 7870 and 7670.  That is unless they do away with 17.3" and go for 18" - that might change things.

[MyPC8MyBrain]

1 hour ago, 1610ftw said:

wait for the new 18" laptop coming out early next year

though this unit is borderline manageable for me moving from 15.6" M3800 im afraid 18" will be bit too big for me,

 

1 hour ago, 1610ftw said:

Your best chance to get the highest level of performance you seek from a Dell laptop

i don't want to convey the wrong message with my efforts, i am not after the highest scores possible,
just want to know i get expected average performance others with same hardware get,
that's the only reason i run benchmark in the first place,

its nice if we could compete at the higher level for the fun of it,

but my only and main goal was to confirm that i invested in the right platform,
a way to validate is to compare results with other similar hardware owners,

 

capping off a 175w sku at 150w advertising that this is the TGP for their sku and then secretly further capping it to 115w is shady practice when they charge an arm and a leg for the sku, far beyond what a new 4090 with external enclosure would cost plus chunk of $400-$500 change left!

to sum i would be most content for the next few years with the 7770 if it could average expected hardware performance, as it is with these shady hidden limitation backed by sophisticated and also shady marketing tactics it falls way behind every other platform with similar hardware, 60w from 115w is over 50% performance cap from the original 175w design for the mobile 3080Ti, the 7770 suppose to be Dell's flagship platform not slackship platform 😞

 

(anyone seeing different values between nvidia system info and cpu-z than shown below?)

[1610ftw]

1 hour ago, MyPC8MyBrain said:

though this unit is borderline manageable for me moving from 15.6" M3800 im afraid 18" will be bit too big for me,

 

i don't want to convey the wrong message with my efforts, i am not after the highest scores possible,
just want to know i get expected average performance others with same hardware get,
that's the only reason i run benchmark in the first place,

its nice if we could compete at the higher level for the fun of it,

but my only and main goal was to confirm that i invested in the right platform,
a way to validate is to compare results with other similar hardware owners,

 

capping off a 175w sku at 150w advertising that this is the TGP for their sku and then secretly further capping it to 115w is shady practice when they charge an arm and a leg for the sku, far beyond what a new 4090 with external enclosure would cost plus chunk of $400-$500 change left!

to sum i would be most content for the next few years with the 7770 if it could average expected hardware performance, as it is with these shady hidden limitation backed by sophisticated and also shady marketing tactics it falls way behind every other platform with similar hardware, 60w from 115w is over 50% performance cap from the original 175w design for the mobile 3080Ti, the 7770 suppose to be Dell's flagship platform not slackship platform 😞

 

(anyone seeing different values between nvidia system info and cpu-z than shown below?)

 

The 3080 Ti is sold with a wide variety of TGPs so actually your performance is probably relatively average, looking at notebookcheck confirms that:

https://www.notebookcheck.net/NVIDIA-GeForce-RTX-3080-Ti-Laptop-GPU-GPU-Benchmarks-and-Specs.588451.0.html

 

Looking only at workstations the 7770 is actually above average so one could also say that in the land of the blind the one-eyed man is king 🙂

 

If you want better you have to get a laptop that will have the TGP you desire and not buy one that hasn't, Dell has these but not in their Precision line.

 

The TGP limitation has also been pointed out early in this thread although I agree that Dell should be more forthcoming with that information instead of giving answers that may give you hope that the lower TGP is somehow a mistake that will be fixed via firmware. MSI and Asus certainly do a better job with letting you know the TGP of their 3080 Ti laptops.

 

As for the 18" unit it is possible that it will not be much bigger than the 7770 if it has a 16:10 screen and if it is too big there is always the option to get one of the 17" units that Dell also sells.

 

 

 

 

 

[Aaron44126]

35 minutes ago, 1610ftw said:

As for the 18" unit it is possible that it will not be much bigger than the 7770 if it has a 16:10 screen and if it is too big there is always the option to get one of the 17" units that Dell also sells.

 

Yeah, I was going to say this as well...  If they do produce an 18" system, it will likely be a similar situation to them moving from 15.6" to 16" — they just made the display 16:10 and kept the width (roughly) the same, so the laptop's footprint is (roughly) the same, but they can market it with a larger display.  There is certainly room in the Precision 7770 to make the display "taller".

 

I saw the rumblings about a possible 18" system but I think that it is likely that Alienware will go first, and next year's Precision "7X80" systems will have the same chassis that this year's had.  (They do consistently tend to use the same chassis for at least two generations.)  Even though Alienware didn't get an Alder Lake HX system, it is looking like they may well get a Raptor Lake HX system (I actually ran across an Alienware guy on LinkedIn who stated he was involved in the project for such a thing), so next year's Alienware lineup might be more interesting.

 

1 hour ago, MyPC8MyBrain said:

(anyone seeing different values between nvidia system info and cpu-z than shown below?)

 

CPU-Z actually shows a blank spot / no value for the GeForce TDP on my system.  Maybe it can't figure it out because I have graphics switching enabled ...?

[Howard N]

Here is a crazy concept if you want to boost the performance. Does anyone offer an external cooling device (much like a micro air conditioner), that can port directly into the input opening vents of the laptop? I think the inputs are under the laptop, so a baffled base might work conveniently. That the fan would be pulling in cold air into the laptop when the fans run, and maybe that would solve the problem at minimal cost?

[MyPC8MyBrain]

10 hours ago, Aaron44126 said:

I have graphics switching enabled ...?

mine is enabled too, it might be our driver differences, i am running latest game ready drivers from NVidia,
you are likely running Dells latest signed drivers? (where Dell has full control on manipulating the values we see!)

 

10 hours ago, 1610ftw said:

looking at notebookcheck confirms that:

way too much info on that page 😄 

from what i can make out there they are not looking at models with 3080Ti rather comparing to other model sku's all together, not one 3080 TGP vs the other or something in these lines but a completely different (earlier or newer) models in comparison to the 3080Ti itself, it doesn't compare how the same hardware fairs in various chassis forms and manufacture designs,

i am looking at stats here for average same hardware scores for comparison,

according to the data there same hardware (12950HX & 3080Ti mobile combo) scores are 

Highest score: 14832  Average score: 13770  Your best score: 12245

the above tells me my system performs 15% below average and roughly 25% below top score,
from the above test and score id consider 100-200 points buffer for variance,
but i cannot account for the remining 2587 points, (mind you that most here don't get near 12k score
its more like 11k average score for Dell 7770 chassis, 2500 points from 12000 is still around 20%), 

 

i am an engineer, most here are also some form of software engineers and various educated professionals utilizing the platform, we work off statistics every day, lets not pretend all of a sudden these statistics don't count, what irks me is our platform is so far off average its not even funny.

if i push a payload that takes a week to compute with above stats
i potentially lose almost two full days working with the 7770 compared to working with other similar platforms, 
we all familiar with the cliché "time is money" it should apply both ways and is very much expected from Dell's flagship workstation platform.

[IamTechknow]

On 12/29/2022 at 3:11 AM, MyPC8MyBrain said:

though this unit is borderline manageable for me moving from 15.6" M3800 im afraid 18" will be bit too big for me,

 

i don't want to convey the wrong message with my efforts, i am not after the highest scores possible,
just want to know i get expected average performance others with same hardware get,
that's the only reason i run benchmark in the first place,

its nice if we could compete at the higher level for the fun of it,

but my only and main goal was to confirm that i invested in the right platform,
a way to validate is to compare results with other similar hardware owners,

 

capping off a 175w sku at 150w advertising that this is the TGP for their sku and then secretly further capping it to 115w is shady practice when they charge an arm and a leg for the sku, far beyond what a new 4090 with external enclosure would cost plus chunk of $400-$500 change left!

to sum i would be most content for the next few years with the 7770 if it could average expected hardware performance, as it is with these shady hidden limitation backed by sophisticated and also shady marketing tactics it falls way behind every other platform with similar hardware, 60w from 115w is over 50% performance cap from the original 175w design for the mobile 3080Ti, the 7770 suppose to be Dell's flagship platform not slackship platform 😞

 

(anyone seeing different values between nvidia system info and cpu-z than shown below?)

I'm curious to know if the Max-Q Technologies being "Yes" accounts for the TDP discrepancy we are seeing here, or has anyone able to rule that out? Max-Q GPUs tend to have much lower power limits, but I think 115W would be too high, I would expect to be seeing an actual TDP of 80W or 90W instead. Or perhaps with dynamic boost enabled, it could be 90W + 25W?

 

EDIT: A day after this post, win32asmguy made a breakthrough that concludes Max-Q Technologies being "Yes" does not influence the power limit.

Edited January 3, 2023 by IamTechknow

[IamTechknow]

4 hours ago, Howard N said:

Here is a crazy concept if you want to boost the performance. Does anyone offer an external cooling device (much like a micro air conditioner), that can port directly into the input opening vents of the laptop? I think the inputs are under the laptop, so a baffled base might work conveniently. That the fan would be pulling in cold air into the laptop when the fans run, and maybe that would solve the problem at minimal cost?

Yes, there is in fact a vacuum cooler that sits next to a laptop's vents to cool it down. Jarrod's tech did a video on this, but he found the difference between the vacuum cooler and a laptop tilted up with a stand/block is minimal:

 

 

[MyPC8MyBrain]

@IamTechknow

we see 115w + 5w boost max

yes it has been verified by at least two separate tests i conducted,

first test was with a Dell 330w power supply with PL4 removed (set to 0 or 330),

second test CPU PL1 and PL2 both limited to 30W leaving additional 80w headroom for the dGpu to use but it refused to utilize 80w available and remain at max 115 while entire system consuming 160w total during bench, 

 

[IamTechknow]

4 minutes ago, MyPC8MyBrain said:

@IamTechknow

we see 115w + 5w boost max

yes it has been verified that the max it will go by at least two separate tests,

 

Gotcha, thanks for sharing your results. Yeah that 120W measurement was no thermal throttling in HWInfo doesn't match up with either the control panel, CPU-Z, or my guess

[MyPC8MyBrain]

if you scroll back few pages you will find detailed posts about these test runs,

[1610ftw]

13 hours ago, MyPC8MyBrain said:

way too much info on that page 😄 

from what i can make out there they are not looking at models with 3080Ti rather comparing to other model sku's all together, not one 3080 TGP vs the other or something in these lines but a completely different (earlier or newer) models in comparison to the 3080Ti itself, it doesn't compare how the same hardware fairs in various chassis forms and manufacture designs,

i am looking at stats here for average same hardware scores for comparison,

according to the data there same hardware (12950HX & 3080Ti mobile combo) scores are 

Highest score: 14832  Average score: 13770  Your best score: 12245

the above tells me my system performs 15% below average and roughly 25% below top score,
from the above test and score id consider 100-200 points buffer for variance,
but i cannot account for the remining 2587 points, (mind you that most here don't get near 12k score
its more like 11k average score for Dell 7770 chassis, 2500 points from 12000 is still around 20%), 

 

Let me help you then with the scores for the 3080 Ti:

 

 

 

And here are all the individual scores for the site that depending on power limit start at very low scores, I have left out the lowest ones. As you can see given its power limit your 7770 performs very well:

 

 

Looking up a benchmarking site where by nature most people will benchmark only when they are interested in benching and when they think they have a good system is let's say a rather weird way of determining what is average especially given the fact that the top results are often achieved not in a normal setting but with some creative manipulation of temperature and/or power limits. @Dell-Mano_Ghas spelled out power limits of this line of laptops early in the thread and it did not look good to me and even then I decided that it would not make sense to even try these out with todays much higher power demands for the new generation of CPUs.

 

This is the last I am going to write about this as you seem to continue to compare your 7770 to other hardware with superior TGP and power limits to conclude that this is something that Dell should offer us. While many here may agree that they should - I am of that opinion myself - they do not and they do not have to.  They sadly do not even have to declare in detail what you can actually expect from their devices which is why we have forums like this one where users and sometimes industry professionals share their results - if in doubt wait and you will know what to expect.

[MyPC8MyBrain]

57 minutes ago, 1610ftw said:

Let me help you then with the scores for the 3080 Ti

Thank you, but common... these are only 41 out of tens of thousands bench results hand picked off the same site i directly compare with, just reposted on their site! their selection also doesn't represent same hardware average but a wide array of different combination and chassis sizes etc., the conclusions you seem to draw are fundamentally incorrect in my view, ill start with Mr. @Dell-Mano_G with all due respects! i never read his posts or visited this site prior to owning the 7X70's, anything he may have said with regards to power limits had no grounds in my decision making or in anyway explains the discrepancy we see atm with declared 150w limit and actual 115w limit,

  

57 minutes ago, 1610ftw said:

early in the thread and it did not look good to me and even then I decided that it would not make sense to even try these out with todays much higher power demands for the new generation of CPUs.

kudos to you for being able to draw the right conclusion and make the right decision to opt out the 7X70 platform beforehand, your opinion about my score is your opinion i have no issues with how you perceive my $4500 investment.

now if you actually studied these scores closely you will also find much higher than 115w sku's with lower score than the 7770 unit gets, that doesn't comfort me or save me the time i expect and paid for, what customers expect in return for their investments is very important to manufactures despite your plight; how we feel about the product should be No.1 concern for any manufacturer planning to stay in business, this is not the army sir. 😊
 

[Howard N]

Thanks for the link to that vacuum cooler video. I was thinking of some kind of device that includes an A/C component so that chilled air would be drawn through the base of the laptop - not just ambient air. I haven't seen anything yet that offers that. I was thinking that a chilled air supply through the base allowing for pulling cold air through the laptop input ports might allow for substantially better cooling if such an option existed.

[Aaron44126]

54 minutes ago, Howard N said:

Thanks for the link to that vacuum cooler video. I was thinking of some kind of device that includes an A/C component so that chilled air would be drawn through the base of the laptop - not just ambient air. I haven't seen anything yet that offers that. I was thinking that a chilled air supply through the base allowing for pulling cold air through the laptop input ports might allow for substantially better cooling if such an option existed.

 

I feel like this is an over-the-top solution that won't help out much in the end.  The laptop's cooling system is more about getting heat from the CPU/GPU out of the system as quickly as possible.  The CPU is going to run at 100°C under full load if the power limits are high enough, and lowering the temperature of the air going in by 10°C or so may help a little bit ... but I have a hard time believing that it would be worth the complexity of setting this thing up plus the power that it would use and noise that it would make just to nudge the CPU speeds up by a few hundred MHz at best.  (Also, would it really be that portable?  Just use a desktop with a bigger cooler...)

 

Plus, for the default configuration, the power limit is more of a performance limiter than the thermal limit is.  (Especially on the GPU side.)  Reducing the input air temperature won't help at all there.

[pintie]

chilled air wouldn't do much.

the bottleneck is the path from the core to air.

i guess you will need 20° K lower ambient temp to get 3-5°C less CPU temperature.

 

 

[MyPC8MyBrain]

a simple device like this goes a long way https://www.amazon.com/dp/B09KXB7QF3
 

[Aaron44126]

Now three times in the last two weeks the system has unexpectedly crashed when I was not using it.  It's been super stable before that.  I have three different BSODs with nothing in common between them.  I'm generally cautious about making system changes so I don't have much to point to.  I am suspecting maybe the 1.8.0 BIOS update?

 

I know they messed with the CPU power stuff in the latest update, so I went and did the RTC Reset procedure to clear the NVRAM to reset everything to "default", to make sure that changes that I made previously (AC/DC loadline fix + unlocking undervolting, which I never actually used) aren't causing the problem.  If it persists, I'll consider rolling back to an earlier BIOS version...

[MyPC8MyBrain]

27 minutes ago, Aaron44126 said:

Now three times in the last two weeks the system has unexpectedly crashed

haven't experienced crashes here yet even with latest 1.8.0, system is on 24/7,
(classic power plan active with S3 enabled via setup_var),

any hints from log viewer?
which apps were running in the background maybe one of them needs an update?
if not disabled you maybe able to find clues in bsod memory dumps,

[win32asmguy]

11 hours ago, Aaron44126 said:

Plus, for the default configuration, the power limit is more of a performance limiter than the thermal limit is.  (Especially on the GPU side.)  Reducing the input air temperature won't help at all there.

 

Tonight I managed to get the 3080Ti in the 7670 working at 125W by using MSI Afterburner to control power limits intead of Nvidia PCF. Here is a 3dmark score:

 

https://www.3dmark.com/3dm/86161666?

 

Basic process, I think:

 

Be on bios 1.8.0, with Nvidia driver 527.56, msi afterburner 4.6.5 beta 2.

In Device Manager, disable "Software Devices -> NVIDIA Platform Controllers and Framework".

Open MSI Afterburner.

The power limit shown will likely be a nonsense value, drag the slider to 1% then back up to 138%, then Apply the changes.

Now the higher GPU power draw over 102W should be observed.

[MyPC8MyBrain]

@win32asmguy 

i think you can squeeze at least 20% more for cpu score,

your dgpu score is as good as i ever seen on the 7670,

are you able to replicate dgpu 125w usage?

HWiNFo will occasionally display odd value calculations
(see below supposedly my dGpu has 750w power limit )

[win32asmguy]

35 minutes ago, MyPC8MyBrain said:

are you able to replicate dgpu 125w usage?

HWiNFo will occasionally have odd value calculations

 

Yes, in Timespy it maxes out at 125W. In games I regularly see 120W+ now where before it was only 102W with the Dell provided Nvidia PCF driver.