Precision 7530 & Precision 7730 owner's thread

My 7730 has a Radeon Pro WX 4150. Is it possible to upgrade that to something that would be more powerful for use primarily with Adobe Lightroom? I spend so many hours each day working in Lightroom that I'm thinking any performance increase would be worthwhile.

 

You could install any of the other GPUs offered in the 7730; Quadro P3200, P4200, or P5200... Finding the card would be the trick. I just searched and found a P4200 on eBay for a pretty penny. I wish eBay sellers would start picking up the term "DGFF" and using it in the auction title...

7740 GPUs would work as well. NVIDIA GPUs would require INF mod to install the driver. (I don't think that this is needed for AMD GPUs.) They'll probably be even more difficult to find, but @Ionising_Radiation got Dell Singapore to send him a RTX 3000 for cheap.

 

One thing I noticed today when testing a Precision 7530 with i7 8750h.

I tested Lenovo P1, Alienware m15, Razer Blade and this Precision 7530. All with the same CPU.

Interestingly, when running prime95 with all of them limited at 35W, the Precision 7530 always runs 300-400mhz lower than the other three.

It seems that something else in the system is saturating the PL budget, leaving less for the CPU.

I doubt it's the memory controller (although 7530 has 2x16gb, the other three have 2x8gb)

Would you guys mind posting your config and your hwinfo DRAM power usage in idle?

It seems at least RAM is indeed using more power than on the other laptops

 

What clocks is the 7530 running at, versus the other three notebooks?

 

Here are some numbers for 7530 (2x16gb) vs m15 (2x4gb) and p1 (1x16gb?can't remember)

2.4ghz vs 2.9-3.0ghz
Yeah it's more like 500-600mhz

IA cores wattage: 31.0-31.1 vs 31.9 (m15) -32.8 (p1)

DRAM with p95 blend: 5.2W vs 2.8 (p1) -3.1W (m15)

I would be inclined to think it's something related to the RAM (such as the mem controller, as the pwr consumption of the RAM itself should not be part of the CPU package power)

 

That's very strange. A 600 MHz difference means that either a) the Precision is the only notebook actually implementing the 35 W power limit, or b) there is something wrong with your RAM configuration. My notebook runs at a flat 2.7 GHz with a 35 W power limit and a -140 mV undervolt.

My DRAM power is at around 1.6 W, and I have three sticks: two 16 GB ECC (which, by all measures, should increase the power drawn), and one 8 GB non-ECC.

 

2.7ghz with 140uv would confirm my data.

Try without uv

Are you getting 1.6W dram usage while p95 is running?
Mine is 1.2w idle, 5.2w with p95 Blend test (first 5 mins)

 

That explains it. The entire point of the Blend test is to hammer the memory controller. My 1.6 W result was with small FFTs.

I still hit 2.7 GHz without the undervolt.

My question is, however: why are you using Prime95, of all benchmarks, to test your notebook? It pits an unrealistically intense workload on a computer, and is only meant to check the very extreme edges of performance. Most notebooks don't fare well with P95...

 

Is the ram in the 7530 single or dual rank? Dual rank memory is more work for the memory controller than single rank memory and can cause higher TDP usage especially in apps like p95.

 

Should the freq not decrease? Wattage should increase with no uv, you should be hitting the PL at lower freq.

About p95: no particular reason, just a reproducible test that I ran on all the laptops. I'm aware it is unrealistic load, but as long as the comparison is fair the result should be reliable enough.
But you are right, I should have gone with small FFT to reduce the impact of the mem controller on freqs. I realized too late and don't have the other laptops anymore unfortunately, so I need to stick to it.
Thankfully, as long as the CPU block is generating the same heat on all laptops, I can still rely on that data to compare the cooling systems.

My main target was to compare cooling and power efficiency in low load states.
To that extent, I am confident the comparison is still fair enough

Thanks for the hint.
Dual rank.
I would be surprised if p1 and m15 were not using the same rank, but yes that might be a possible cause