Having confusing desktop GPUs wasn't enough - Nvidia removing MaxQ branding

[linuxChips2600]

23 minutes ago, Bombastinator said:

What do you mean “theoretically”. That has ALWAYS been the case.

Linus himself explained in a recent WAN show that part of the "hype" around the gtx 900 (i.e. Maxwell) series of GPUs was that finally desktop and laptop graphics mobile performance have both reached a point to where there was supposed to be "minimal" performance differences between the two, which is why he was all the more surprised when there was info leaked about mobile RTX 3000 series having performance much closer to Turing mobile rather than desktop Ampere, and that info leak was before this news about Nvidia dropping "official Max-Q branding".

[Bombastinator]

3 hours ago, linuxChips2600 said:

Linus himself explained in a recent WAN show that part of the "hype" around the gtx 900 (i.e. Maxwell) series of GPUs was that finally desktop and laptop graphics mobile performance have both reached a point to where there was supposed to be "minimal" performance differences between the two, which is why he was all the more surprised when there was info leaked about mobile RTX 3000 series having performance much closer to Turing mobile rather than desktop Ampere, and that info leak was before this news about Nvidia dropping "official Max-Q branding".

It’s all about the batteries.  That might have been true for a short time some years ago.  There was a period where GPUs got radically more power efficient, and the standard for gaming was 1920x1080@60fps.  A 1650 will push that fine.  It fits in the power envelope.  They’ve done tests on equivalently named GPUs. They are not equal.  Why would they possibly be?  One needs 200w to run and the other uses a tenth of that.  A card using that little power on a desktop machine could be made half height single slot and would need no additional power connectors. Why on earth would they not MAKE them that way if they could? There are a lot of people with business prebuilts that would be crying for even 2060 performances in a no power connector card needed even if it was full height double slot.  Half height would double the resale value of some desktops. The best that can be gotten though is a 1650.  Even a 1650ti draws too much power.  That enthusiast GPUs simply waste hundreds of watts for no good reason is magical thinking.

Edited January 23, 2021 by Bombastinator

[CephDigital]

4 hours ago, Bombastinator said:

What do you mean “theoretically”. That has ALWAYS been the case.  If you buy a laptop with an “X” gpu in it it is going to be a ton slower that that gpu as a desktop version.  The whole problem is calling laptop GPUs by desktop gpu designations in the first place.  They used to not even have the same gpu chip. They would just “call” a given laptop gpu by a desktop gpu name to designate its pricing level.  They had nothing to do with each other except marketing shenanigans.  Lately they’ve actually been putting the same chip in the laptops but running it at a lower wattage.  A 2080 or whatever, in a laptop never behaved at a desktop 2080 level.  This is why the “M” was a big deal.  The mobile version is ALWAYS a much lower performance version.  If you’ve got a chip that runs at 200w and needs 3 pounds of aluminum fins and heat pipes to cool it thinking that putting the same gpu in a laptop will net the same performance when it’s running under 20w with maybe a vapor plate and a tiny little fan for cooling will perform on the same level is stupid.  People seem to do it though. A maxQ 2080 or maxP 2080 won’t run at normal 2080 speeds.  You could call it an 8gbtu102, with some or another power level, because that’s what it is, but it was never a 2080. The Q in maxQ stood for “quiet” iirc whereas the maxP stood for “power” it was never Max though.  It’s all just stuff tarted up with marketing.  

This isn't about mobile vs desktop but purely on the mobile side. It's to be expected that mobile wouldn't perform as good as desktop. Expecting otherwise would mesn you had a VERY chunky "laptop".

 

The fact that they HAD a system in place to somewhat tell how good a GPU between laptops would be and then removing it is just plain misleading.

 

MaxQ would tell you it priorities silence and size.

MaxP would tell you it's more for cramming as much power as possible into a laptop.

[Speedbird]

22 hours ago, StDragon said:

It's more rare than not, but low-end desktops and high-end laptops do cross over. For example (don't know if that's now true), but some Dell Precision laptops in the past used full Intel socketed Xeons whereas all other laptops used mobile CPUs. In the case of Dell OptiRimo desktops, they all use desktops CPUs whereas the OptiRimo Micro series will use a mobile CPU. The difference is often a combination of core count, speed, and/or lower power consumption.

 

Desktop example: Dell OptiRimo 7080 uses the Core i5-10500 whereas the 7080 Micro uses the Core i5-10500T. Click on the links to see the differences.

 

And yes, they are distinguished by the "T", but I doubt most consumers pay attention.

T-class CPUs are still desktop CPUs that use the same socket. They only differ from non-T parts in TDP (35W) and a lower clockspeed. This is nothing like the H vs U CPUs on mobile, which can have very different core counts.

[Bombastinator]

11 hours ago, CephDigital said:

This isn't about mobile vs desktop but purely on the mobile side. It's to be expected that mobile wouldn't perform as good as desktop. Expecting otherwise would mesn you had a VERY chunky "laptop".

 

The fact that they HAD a system in place to somewhat tell how good a GPU between laptops would be and then removing it is just plain misleading.

 

MaxQ would tell you it priorities silence and size.

MaxP would tell you it's more for cramming as much power as possible into a laptop.

There is power budget issues in laptops everything is always “maxP” and always has been.  You mean like the “m” designation?  The reason that got removed I think is the whole thing with laptops is the companies put a lot of marketing money into building brand recognition for their desktop video cards.  Various desktop cards get known by their number designation as able to be able to do “x” with a game.  Marketers attempt to leverage this by naming their laptop video solutions after their desktop video solutions even though the two are not even remotely comparable. What one used to get from nvidia as example was “(number)M”. The “M” meant “that number you just read? This card has nothing to do with that one”.  Customers slowly twigged to it.  So they changed “m” to “maxQ”. In the great scheme of things nothing really changed.  There is a small bit more resemblance with maxQ over M in that the maxQ video solutions (because you can’t call them cards) actually generally have the same GPU chip and ram type from the desktop version.   Functionally though it’s exactly the same problem.  

[Bombastinator]

11 hours ago, Speedbird said:

T-class CPUs are still desktop CPUs that use the same socket. They only differ from non-T parts in TDP (35W) and a lower clockspeed. This is nothing like the H vs U CPUs on mobile, which can have very different core counts.

So? The still don’t run the same.  The problem is marketers are attempting find a way to attach the desktop identifier to the laptop video system in order to make it seem faster than it is.  A 2080maxQ is still around 1660 performance because of the low wattage. What might be interesting is doing a high end gpu at low wattage making low heat output as a half height single slot card. You’d get a $600 1660, which would seem awful, but it would fit into a half height business prebuilt and be the fastest card of its class.  People seem to treat the “why does the 1650 sell so well?!” as being this inexplicable mysterious thing.  It doesn’t seem that way to me. It’s the fastest card in its class of “fits inside an office machine”   People get their office machine as a perk for home use and want to play games on it but the biggest card that will fit is a 1650 which IS fast enough to play games at 1920x1080@60fps. So the buy one.  There are a LOT of machines like that, so a lot of 1650s get sold.  If AMD wanted to shoot Nvidia’s 1650 sales in the face, they could do a low power 5700xt that would be a bit faster than a 1650 but still fit in the heat/power envelope needed for those machines.  Then the 1650 would no longer be  best in class.  Sure they wouldn’t actually SELL very many of em, but the would smear the 1650 halo. AMD doesn’t even try to play marketing games with their laptop video solutions though. They have a whole other designation system.

Edited January 24, 2021 by Bombastinator

[PianoPlayer88Key]

My personal opinion...

 

I think if a video card (laptop or whatever) cannot match the performance of at least a reference / stock 3080 - should NOT be called a 3080, or 3080 MaxQ, or 3080M, or whatever.  If it matches, say, a 3060 Ti, then CALL IT a 3060 Ti.

(I wouldn't mind a designation for video cards that are binned for better power efficiency, though.)

 

Same goes for desktop vs laptop CPUs.  I don't like seeing situations where a ULV mobile (<6W TDP) Core i7 or Ryzen 7 is outclassed by a desktop Core i3-K or Ryzen 3-X of the same generation.

I prefer that within the same generation, if a mobile CPU is not at least as fast as a full desktop Core i5 or Ryzen 5, then it should be called a Core i3 / Ryzen 3, Pentium / Celeron, Athlon, or something. 

Maybe another branding, like they already do with servers (Xeon/Epyc)?

 

 

 

Someone earlier in the thread I think mentioned laptops with desktop CPUs in them.

My laptop has an i7-6700K, and the same company a few years before my laptop's generation had a laptop with a Sandy/Ivy Bridge LGA 2011 socket.

[Junaid Hameed]

On 1/22/2021 at 5:07 AM, dizmo said:

Not entirely true, as @IAmAndre said.

It still doesn't change the fact that anyone buying something that expensive should do their due diligence and research what they're buying.

That's possible ONLY if the manufacturer is willing to tell us. Not many are that tech savvy to read "between the lines"

[Speedbird]

2 hours ago, PianoPlayer88Key said:

My personal opinion...

 

I think if a video card (laptop or whatever) cannot match the performance of at least a reference / stock 3080 - should NOT be called a 3080, or 3080 MaxQ, or 3080M, or whatever.  If it matches, say, a 3060 Ti, then CALL IT a 3060 Ti.

(I wouldn't mind a designation for video cards that are binned for better power efficiency, though.)

 

Same goes for desktop vs laptop CPUs.  I don't like seeing situations where a ULV mobile (<6W TDP) Core i7 or Ryzen 7 is outclassed by a desktop Core i3-K or Ryzen 3-X of the same generation.

I prefer that within the same generation, if a mobile CPU is not at least as fast as a full desktop Core i5 or Ryzen 5, then it should be called a Core i3 / Ryzen 3, Pentium / Celeron, Athlon, or something. 

Maybe another branding, like they already do with servers (Xeon/Epyc)?

 

 

 

Someone earlier in the thread I think mentioned laptops with desktop CPUs in them.

My laptop has an i7-6700K, and the same company a few years before my laptop's generation had a laptop with a Sandy/Ivy Bridge LGA 2011 socket.

Not going to happen unfortunately. So many people just have a philosophy of "bigger number better" when buying a computer. 

[PianoPlayer88Key]

1 hour ago, Speedbird said:

Not going to happen unfortunately. So many people just have a philosophy of "bigger number better" when buying a computer. 

yeah ... I want that to ACTUALLY be true, though, as in,

"ALL 11th-gen Core i5's are faster than ALL 11th-gen Core i3's," and

"ALL 12th-gen CPUs are faster than ALL 11th-gen CPUs."
Or,

"ALL RTX 4070s are faster than ALL RTX 4060 Ti's," and
"ALL Hopper GPUs (or whatever architecture is next) are faster than ALL Ampere GPUs."

(The only overlaps, and that only MAYBE, might be overclocking under LHe in Antarctica in the dead of winter.)

[dizmo]

9 hours ago, Junaid Hameed said:

That's possible ONLY if the manufacturer is willing to tell us. Not many are that tech savvy to read "between the lines"

So then buy from someone else? You're making it seem like there's only one manufacturer of laptops. There isn't. There's many.

[Mark Kaine]

On 1/22/2021 at 2:31 AM, TrigrH said:

Even the GA104 (3070 die) runs fairly hot and pulls a good chunk of power (close to the 2080 if i'm not mistaken).

depends on the model perhaps but generally a 3070 will draw more (mine is at 270w continuous, 300w max for example) its a bigger chip too afaik so it makes sense. 

[Junaid Hameed]

12 hours ago, dizmo said:

So then buy from someone else? You're making it seem like there's only one manufacturer of laptops. There isn't. There's many.

Yes. But if Nvidia starts giving leverage, all will grab it. Soon all will follow the same misleading marketing pitch. Example : recent price hike of GPUs. tariff was for US only. they increased prices worldwide.  

[CTR640]

On 1/22/2021 at 12:51 AM, dizmo said:

Doesn't seem to be too different than mobile CPUs, they too can be had in different configurations.

While it might be a little backhanded, and it'd probably be better if they didn't, I don't fault them fully.

If you're a consumer, and you're spending over $500 on something, you should do your due diligence and put in an hour or two of research.

And what if (web)stores are not putting information like which GPU their laptops have?

Some stores just put like only GTX950M and not wether it's DDR3 or GDDR5 version. Is it still customers their fault?