7540 w/ RTX 4000 & 5000

Basically yes...
The RTX 4000 for desktops and RTX 4000 for mobile share the same name, but they don't share the same chip, the RTX 4000 in desktops actually has less CUDA cores than the mobile one (but will still perform faster thanks to the higher power limit).

On the mobile side, all of the cards labeled "RTX 4000" have the same GPU chip. Each manufacturer basically sets a power limit for the chip for each model laptop that they want to drop it in. I haven't seen any of them actually explicitly labeling their RTX card "Max-Q" but in 15" systems it is normally capped at 70W or 80W. A lower power limit translates to a lower max clock speed for a given workload, which translates to lower performance.

(Haven't seen any benchmarks for RTX 4000 in 7540 yet. With the power limit it might not be that much faster than the RTX 3000...)

7740 has RTX 4000 capped at 100W (or maybe just a little bit higher), basically the highest limit that you will find for this card.

 

Great, just the info I was after.

Thanks a lot.

It's interesting - without doing the research, it's not at all apparent to the average consumer that the same card in a 15" will run slower than in a 17".

 

And that's the entire point of the 'Max-Q' buzzword. It is to sell the idea that you can stuff the performance you can get out of this card:
In a notebook as thin as this:


This is physically impossible: the power budget of the desktop card is greater than that of the entire notebook. Hence NVIDIA, and other OEMs have quartered the power budget available, and they tend to be on the conservative side—in other words, even if the cooling in the notebook can handle it, clocks are likely to be lower to allow some headroom.

 

I get that it's obvious you're not getting a desktop card in a laptop, what i mean is that if you were choosing between a 15" and a 17" laptop which each had an RTX 4000 listed as the GPU, it's not obvious that the one in the 15" would be less powerful than the one in the 17".

 

Ah, I missed out another point I wished to add. There was some open branding of Max-Q with Pascal-generation cards, but the fact remains that OEMs are able to adjust and lock power limits as they see fit.

The Quadro P3200 in the 7530 was around 10-12% slower than the P3200 in the 7730 because the former had a 65 W power limit, whereas the latter had an 88 W power limit. Cross-flashing VBIOSes closed the gap, and there was no obvious reason why a slower VBIOS was allocated to the P3200 in the 7530, because there was plenty of thermal headroom.

The exact reasons for these are known only to people privy to Dell's engineering, marketing and finance teams.

The average consumer, as you've mentioned, sees the number ('RTX 2080' or 'Quadro RTX 5000'—though much less of the latter), and assumes they get desktop performance, and the companies get a sale. It's an easy marketing tactic.

 

Yep. Although the RTX 4000 in the 7740 is a little below the desktop RTX 4000 performance, it seems that it'll be a pretty decent GPU in a laptop (without spending a huge amount of extra money to go up to the 5000).

I'll be using it for CAD work so want a Quadro rather than a GeForce.

 

We do have RTX 4000 vs RTX 5000 (in 7740) benchmarks. The RTX 5000 (mobile) has the same chip as the desktop RTX 5000 (230W) and GeForce 2080 SUPER (250W), but the RTX 5000 in the Precision 7740 has the same 100(/110?)W power cap as the RTX 4000. Because of this, the performance gain of selecting the RTX 5000 over the RTX 4000 is minimal; it is perhaps around 5% faster with most workloads, a hard sell given that it costs around $1,000 more than the RTX 4000.

I prefer how things were pre-Pascal when NVIDIA gave their mobile chips different names (the "M" suffix) to make it clear that they were different.

 

Thanks, that's interesting. I'm just about to place the order for a 7740 with the RTX 4000. I guess the 5000 might be worth it only if I needed the extra 8GB VRAM, which I doubt I do...