Precision 7550 & 7750 Owners' Thread

Thank you all for your precious advise!

I've ordered 7750 with i7-10875/8GB RAM/256GB SSD/RTX4000/ Full HD with IR camera.
I'll be adding 4x32GB HyperX Impact 2933 CL17's and 2TB ADATA SX8200Pro.
Now the worst part: 5 weeks of waiting time...

I have an idea:
You can charge 7750 via USB-C/TB3. It takes 130W per port. So at least in theory I could use Omega 200 charger to have 200g power supply instead of Dell's 850g one.
That could be nice weight saving for 90USD.

 

Interesting idea, and I'm curious to see how it works out...
But more than likely, it will work OK for charging but the system will throttle pretty hard if you're not using a Dell power supply.

 

You can charge 7750 from WD19DC and then it uses Power Delivery via TB3. So it shouldn't throttle. Worst case scenario - it'll consume some of the battery when crunching heavy workloads (I don't think it's possible to use sustained 200W of power under normal cirucmstances).

 

Hi there @ov_Darkness,
The 7750 will only be able to pull between 90W & 100W from a non-Dell type C power adapter. Pulling 130W from a type C adapter requires a Dell branded adapter.
On the 7750 anything less then 210W of power from a power supply will cause the system to throttle.
Tks
Dell-Mano_G

 

Following up on my "prediction" here... as multiple sources are today reporting details of the GeForce 3060, 3070, and 3080 mobile lineup. Interestingly, the mobile GeForce 3080 is listed as having a GA104 chip and the maximum amount of CUDA cores for that chip, which is 6,144. (Only a modest boost over the GeForce 3070's 5,888 codes... though some sources are reporting 16GB vRAM for the mobile 3080.) The desktop GeForce 3080 has a GA102 chip and 8,704 CUDA cores. That will mean a non-trivial performance difference between the two GPUs (power/thermal limitations aside). And, if these specs are right, it would be the first time that they've differentiated the specs (other than power limit) of the same-named desktop and mobile GPUs since they dropped the "M" suffix with the Pascal series.

Anyway, the mobile 3080 is looking a lot like what I have speculated for the mobile RTX A5000. I'll be rather surprised if we see more than 6,144 CUDA cores in the 2021 Precision top GPU card.

 

Thanks for clarification!
So two type-c 100W connections won't make it
I was hoping to get Omega 200, because this Dell branded 240W PSU is MASSIVE and I travel with my laptops a lot.
Any chances for Dell GaN PSU anytime soon?

 

@Aaron44126 Obvious question: will they fit previous Precision generations, or their TDP will be so high there will be no way to do that?

 

I strongly suspect that the GPU TDP will be the same (90W in 15" / 110W in 17"). In the past few generations, Dell has made thermal improvements but they have mostly been used to shrink the system. Previous comments from Dell indicate that they're not interested in pushing the power envelope any higher. Doing so would require making the system larger which they do not have an appetite for. So, performance improvements must come from improved power efficiency and process shrinks (from NVIDIA) and not just gobbling up more power. (Turing to Ampere is 12nm→8nm so there will definitely be a power efficiency gain there. Tiger Lake will also be the most interesting CPU upgrade from Intel in many years.)

Aside — I personally think that there may be room to push the GPU up to 120W or higher for some workloads... current NVIDIA GPUs seem to power throttle well before they thermal throttle in these systems. Unfortunately, no one has found a way to override the power limit coming from the NVIDIA GPU vBIOS to try it out. Dell recently pushed out a vBIOS update for high-end cards in the 7550, raising the power limit from 80W to 90W, so they're at least thinking about the possibilities. (FWIW, I understand the desire for larger more powerful systems but I think that Dell has struck a good balance between power and size for a high-end business system. They would probably lose sales if they made it larger or gave it giant rear exhaust blades like you see in some gaming laptops, and there's no sense releasing a special separate larger system to cater to a niche market. Not to mention, if you're complaining about lugging around a 240W power brick, think about what a 300+W brink would be like...)

There's a decent chance that you will be able to use an Ampere NVIDIA GPU from the upcoming generation (7560/7760?) in the current generation (7550/7750). We won't know for sure until we get a look at the next-gen systems and confirm that the card form factor is the same... and then until someone actually tries it and confirms that there are no issues (i.e. BIOS conflict). (7X60 systems with 11th gen Tiger Lake CPUs will probably be revealed in May at the earliest, but there's always a chance for leaks.) No hope for systems older than 7X50 will not work because Dell changed the GPU card shape.

There is precedent for this type of upgrade; for example, it is possible to take a Turing GPU from a 7X40 system and install it in a 7X30 system, because the GPU cards for those systems have the same form factor and interface. Actually getting your hands on a GPU card to perform the upgrade can be challenging and expensive. Dell will not sell one to you standalone in most territories. RTX 5000 cards are selling on eBay for over $1,000 (though... I think that they were up around $2,000 when they first appeared). And with the general upward trend in GPU prices, and the supply issues that NVIDIA has been having... I wouldn't be surprised if the price to add a RTX A5000 card to a Precision 7760 tops $2,500 at launch.

 

Which 240W brick do you have right now? The rounded-edge one (DPN 7XCR6) that they have been using is not that bad.

 

Continuing my unrelated self-conversation here...
Specs for mobile GeForce 3070 and 3060 coming in, with 5,120 and 3,072 CUDA cores, respectively... in each case lower than their desktop counterparts (5,888 and 3,584)... Yuck, NVIDIA. Anyway, this gives a further look at what we might expect to see with the next-gen Precision GPUs. RTX A4000 is likely to look similar to the (mobile) GeForce 3070, with maybe a bump in vRAM.

No doubt these GPUs are a good upgrade over the mobile 2000-series, the yucky thing is if they keep naming them the same as the desktop versions, "pretending" that they are the same while keeping the blatantly different specs. (Though in the end it might not really matter that much, in terms of performance, because GPUs with more CUDA cores would be pretty severely power limited anyway... Basically, the power limit determines performance, not the number of cores, once the number of cores is high enough.)