Precision 7530 & Precision 7730 owner's thread

So I got my thermal paste today, decided to have some fun. I also undervolted by 0.050 in XTU. Here's what I experienced:

Fire Strike: No change. 11,150-11,250. Definitely still power limited
Cinebench runs: 1250 first run, 1180-1190 on subsequent runs. A marked improvement over 1150-ish! However, this too is power limited to just 60W, as CR3 said about the 7730.

As for temps, they seem a smidge lower overall, if any. GPU never exceeded 62c during Fire Strike, which is cool, but CPU still gets toasty. It hit a max of 95c during the Cinebench runs, though it never thermal throttled.

Bonus photos:

@Ionising_Radiation here's that heatsink picture, finally.


And the paste application. I tried being less generous with the CPU and it didn't like that too much. Using more greatly improved the situation.


While I was in there I added a piece of foam underneath the middle trackpad button to help absorb some of the vibrations from the speaker. I did a similar mod to my 7510 and it really helped reduce the distortion rattling at higher volumes. So far it seems to be working well here too!

 

Must've been a pain to disassemble due to the inverted mobo.

Why only -50mV? Coffee Lake should be able to undervolt much more than that, at least -100mV (make sure you undervolt both core and cache equally).

What paste did you use, NT-H1? I haven't heard many good things about it in notebooks, hope it doesn't pump out on you.

 

I'll give -100mV a shot. I suppose I was being overly cautious/lazy heh. The paste is actually Cooler Master MasterGel Maker Nano.

Edit: -100mV scores 1190-1200 on the 2nd run and beyond, neat!

 

Just some advice:
- -130 MV should be no problem at all for your CPU.
there is absolutely no harm done by too much under-volt, so test away. just harmless Blue Screen and reset.

-That is WAYYYY too much thermal paste, especially on CPU. this should be about max for reference:

 

Wanted to get your guy’s thoughts on a computer purchase. Been looking at the Precision 7730 with i9 processor and P5200 GPU. I mainly work video edit / vfx (Premiere, After Effects, Resolve, Fusion, etc). Secondary would be gaming. Looking for a desktop replacement laptop. Was wondering if anyone had thoughts regarding resolve performance using a P5200 Max-Q in the Dell Precision 7730 vs GTX1080 say from the Alienware R5 17. Would I see significant improvement in the Alienware over the Precision 7730? Is the 16GB vRam vs 8GB vRam important?

 

Yeah you should be able to do much more. I would say straight jump to -100mV and lower from there, but then again going step by step isn't a bad idea either. The most important thing you gotta watch out for when undervolting using an offset in adaptive voltage mode is freezes/BSODs when the laptop is idling, as well as non-crash instability (WHEA correctable errors in the Event Viewer system log). The offset applies to the entire voltage range, so what may be stable at a higher voltage under load may not be stable when the voltage drops during idle.

Good choice for the thermal paste.

Putting too much paste isn't a problem (if it's non-conductive stuff), the excess just gets squeezed out over the sides. Putting too little can result in higher temps as @Div033 mentioned, since most laptop heastinks are low pressure.

Honestly that looks like too little paste to me, especially on the GPU, unless you have very flat heatsinks and good pressure.

 

-----------
XeonPlanner said: ↑
Link is here: https://forum.51nb.com/forum.php?mod=viewthread&tid=1845538&extra=page=1
I found out that P5200 is limited to 110w, a typical value for 1070 instead of 1080.

---- you said,
I was seeing 125-130w on the P5200. Over 200w at the wall.
-----------

so I think you looked at my review in that 51nb link, and looked at my photos and you make a conclusion: I was seeing 125-130w on the P5200. Over 200w at the wall.
You conclusion, again, I don't mean to offend you, is really incorrect. I say it again, P5200 on 7730 cannot sustain above 125w at all. It is around 110w at most. You can test it on your system 2 as you said.



By the way, I could be wrong but based on my experience, I think XTU is almost useless on 7730, you can also check here, https://forum.51nb.com/forum.php?mod=viewthread&tid=1845119&extra=page=1
that guy is 8950+P5200 as I said, and he tried to modified the settings in XTU. NO use at all.

Peak power is locked at 90w, for the first 30s.
Then regular power is 60w.
This is actually locked by dell. I say it again, it is locked.

No matter how he modified the power and peak power during time or something (28s change to 96s), it actually has no effects at all. like I said 90w for firsr 30s, then 60w. This cannot be changed. this is locked by dell.

And I can confirm XTU is not useful for 2186M+P5200. The power behavior is locked by Dell. You can see that you can 'adjust' them on XTU, But, as I said, no use really. it is locked, 90w for first 30s, then 60w constant.

As to the frequency you mentioned, again, I don't mean to offend you but, I don't think it is really useful.
1. as long as FPU(AVX) is involved, your frequency cannot be 5.2ghz. Even if you have 5.2ghz, it is just the result of burning pure cpu and is just a single core, not FPU with AVX. I guess here we don't care too much this. If you burn FPU along, it is only 3.3Ghz or so. Nowadays, actually FPU score is the thing that is really useful in real use. Cannot imagine your program is AVX irrelevant.
2. and as long as power is locked, the performance is basically locked. it doesn't really make any difference whether you overclock it or not. If you really want to overclock, you have to unlock the power limit first. For now, it is impossible. Unless you have ways to directly modify Dell's bios.

Yes sure you can offset some voltage or something, but if power is actually locked, I feel you just cannot expect too much from those voltage offset things.

 

The AW 17R4 GTX 1080 scores 35-40% higher in Fire Strike Graphics than the 7730 P5200 Max-Q:

Spoiler: GTX 1080

Spoiler: P5200 MQ (stock paste)

Spoiler: P5200 MQ (repasted)

 

I agree on the 5.2ghz part. Not really useful, was just surprised that it got over 5.0ghz. 5 seems pretty stable in the AW17 and I would probably use the same setting in XTU on the 7530/7730 if I had an i9 version.

On the other, I was relaying what I was seeing on the P5200 system I was testing in my office. 125w on the gpu, 199-200 max (original post was cut off here) 199-200 max under total system power, and 215-221w at the wall on the Kill A Watt meter.

I know this is not what other people are seeing on the GPU, I am trying to figure out why.

 

I don't know man. I have a ton of experience with experimenting with this. Had over 25 laptops; always had more success with the less/ grain of rice method. Sooooo many debates over this though; with little to no conclusion. i would suggest @Div033 try both... can't hurt.