Nvidia GeForce RTX 3070 mobile GPU in upcoming HP Omen 15 gaming laptop offers 36.9-49.3% performance improvement over RTX 2070 Max-Q

[Redaktion]

An Nvidia GeForce RTX 3070 mobile GPU has been put through its paces on Geekbench as part of an HP Omen 15 gaming laptop. The RTX 3070 delivered high OpenCL scores that place it up to 49% faster than a GeForce RTX 2070 Max-Q and even at a similar processing level as the desktop variant of the RTX 2080.

https://www.notebookcheck.net/Nvidia-GeForce-RTX-3070-mobile-GPU-in-upcoming-HP-Omen-15-gaming-laptop-offers-36-9-49-3-performance-improvement-over-RTX-2070-Max-Q.512632.0.html

[Gigaboly]

Disappointed as expected... need to compare them to super cards not regular maxq, if 3070 will be 36-49% faster than 2070 maxq than maybe 20-30% faster than 2070 super, which is not that much jump compare to desktop, especially if the price will go over 2k

[Warney]

Get wrecked, the article says that it was a max Q variant of the 3070, delivering 2080 desktop performance. If you don't think this is a massive generational leap you don't know what you're talking about.

I'll pay 2k plus for that.

Disappointed as expected... need to compare them to super cards not regular maxq, if 3070 will be 36-49% faster than 2070 maxq than maybe 20-30% faster than 2070 super, which is not that much jump compare to desktop, especially if the price will go over 2k

[Owen]

so rtx2060 level performance in a 13-14 inch laptop is possible? Man, I will gladly pay $2000 for a 14" with 5800u (or 5800H?) and rtx 3060 (or maybe rtx 3050?). The portability and performance is just too good for my line of work.

[S.Yu]

From the looks of it it's a cut down 3070, still wider than a 3060Ti? One really has to wonder how the desktop version could have such high TGP while the laptop version could cut that by so much and remain in line with the last generation.

[KumaHIME]

From the looks of it it's a cut down 3070, still wider than a 3060Ti? One really has to wonder how the desktop version could have such high TGP while the laptop version could cut that by so much and remain in line with the last generation.

The desktop version has such a high tdp because nvidia tried pushing the clock speeds as high as they could. If you reduce clocks from 2GHz to 1.9GHz, you can take off some 75W of max power consumption off. You have to keep in mind that clock speed and power consumption do not scale proportionally. i think it's more of a quadratic curve that relates the 2. A personal example i can give is for my desktop rtx 2080. if i let it go to 1930MHz, it will pull 230W. it pulls 175W @1800MHz, 130W@1700MHz, and 100W@1600MHz. so you can see lowering even just 100MHz can greatly reduce power draw. comparing min and max from my data set, you can get 83% of the performance (theoretically) while drawing only 43% of the power that you would at max. this is also why nvidia pushes for multiple tier maxq gpus even though power consumption is the same throughout that lineup. by having more cores (adding cores increases power consumption more linearly), but lower clocks, they can achieve "noticeably" better performance with similar power draw.

[Jd1982]

The "Buy The Omen 15 on Amazon" link doesn't take is to a page where we can purchase the laptop in the article (with the RTX 3070). The link goes right to an Omen with the 2060 instead.

[S.Yu]

From the looks of it it's a cut down 3070, still wider than a 3060Ti? One really has to wonder how the desktop version could have such high TGP while the laptop version could cut that by so much and remain in line with the last generation.

The desktop version has such a high tdp because nvidia tried pushing the clock speeds as high as they could. If you reduce clocks from 2GHz to 1.9GHz, you can take off some 75W of max power consumption off. You have to keep in mind that clock speed and power consumption do not scale proportionally. i think it's more of a quadratic curve that relates the 2. A personal example i can give is for my desktop rtx 2080. if i let it go to 1930MHz, it will pull 230W. it pulls 175W @1800MHz, 130W@1700MHz, and 100W@1600MHz. so you can see lowering even just 100MHz can greatly reduce power draw. comparing min and max from my data set, you can get 83% of the performance (theoretically) while drawing only 43% of the power that you would at max. this is also why nvidia pushes for multiple tier maxq gpus even though power consumption is the same throughout that lineup. by having more cores (adding cores increases power consumption more linearly), but lower clocks, they can achieve "noticeably" better performance with similar power draw.

Thanks, eye opening. I know there's a different efficiency curve for each component but the slope here is quite a bit steeper than I thought.
5120 vs. 5888 still means that they've cut down part of the silicon though.

[Gigaboly]

Warney  -  desktop 2080 is 215W of power, 3070 maxq will be 80-90W, there is no way they come close, If you think this is a massive generational leap you don't know what you're talking about.

[Jose Flores]

I scored 103678 in my Alienware m15 R3 with a 2070 Super (non Max-q) at 2085mhz in core and 1720mhz in memory

[Ryguy]

Not bad but still not beating my 2080Super which I scored a 127,054. And this is even in a laptop mind you(Area51m r2) lol. Understand the 3070 card needs 200+ watts to perform, good luck making that work in anything less than a desktop replacement laptop lol. Technology isn't there yet. This is why they do a max-q variant. The wattage is reduced by more than half(ouch) to be able to stay cool, and work in regular as well as thin&light gaming laptop chassis'. This is especially done in chassis' that share the same cooling for both the cpu and gpu. Even if\when a Non max-q 3070|3080 comes along it will still be limited in power also(compared to it's desktop variant), bc it will be a "mobile" version. The only way you will see actual "Desktop" card like performance is with a "Desktop" replacement laptop, bc the chassis on them can handle the heat. However there are currently only just a few companies I know who can achieve this. One of them being Alienware bc they use their own propieraty mxm style graphic  chips. I guess what I'm trying to say is don't expect to see crazy gains unless you opt for a desktop replacement laptop. Sure a 3070 max q should be better than a 2070 max q, however since both are limited to roughly 90watts I wouldn't expect crazy gains. Sure in this test the benchmark shows the 3070 crushing the 2070 by what was it some 40%? Thats great and all but as history has shown real world gaming results are much different synthetic benchmarks. For instance if I overclock my 2080Super an easy +150core|500mem I achieve a score of 127,550 which crushes the 119,479 3060ti score. However though in real world gaming both cards trade blows depending on the game, yet you would think the 2080Super should do better bc the 3060ti didn't come close in the benchmark scores.

[Eduard Caro]

The RTX 2070 Super Mobile is already around 25% improvement in performance over a RTX 2070 Max-Q. So it is still a good improvement but not that crazy.

[benjibear]

The "Buy The Omen 15 on Amazon" link doesn't take is to a page where we can purchase the laptop in the article (with the RTX 3070). The link goes right to an Omen with the 2060 instead.

It's not out yet silly.  ::) The GPUs were just announced this week. Everything is pre-production/pre-orders.

[Gigaboly]

3070 mobile has 88000 point only, while 2070 super maxq is at 79000, so the improvement can be around 10fps in games for a lot more in price