The Notebook Review forums were hosted by TechTarget, who shut down them down on January 31, 2022. This static read-only archive was pulled by NBR forum users between January 20 and January 31, 2022, in an effort to make sure that the valuable technical information that had been posted on the forums is preserved. For current discussions, many NBR forum users moved over to NotebookTalk.net after the shutdown.
With the recent news that desktop GTX 980 with 165W and all the way up to 200W (Clevo) is getting ready for notebooks, I can`t stop but wonder where the hell AMD are?
Why not launch R9 Nano and introduce HBM for the very first time for notebooks? TDP for Nano is 175W and the performance is equal to GTX 980 in 1080p and 10 % faster in 4K.
Which means same level as Nvidia in 1080p notebooks and better in those with 4K displays.
Perfect opportunity to get back on the race again. But AMD is nowhere to be seen...
To fit an HBM chip on MXM module it needs to be redesigned. First the spec, then the actual module (which follows the spec). By standard MXM has size, hole spacing and placement, chip placement, vRAM spacing and placement, VRM placement and etc, you get the picture. A guide, if you want, how the MXM module should look and act like. Current HBM chips (and I would guess the ones in future as well) are pretty big, and they can't fit between the current MXM 3 mounting holes, hence why it needs a redesign. It would fit in the current MXM-B format, it just needs wider hole spacing and there would be space for like 6 phases, which should be enough. Now, the interesting part. This new 980 module is obviously new and nothing we have seen so far... unless ASUS. It seems that more than one brand would have a notebook with it, so it's not proprietary, but rather standard. I'll leave the conclusions to you. MXM Sig is owned by nGreedia, to enhance your guessing experience. AMD is a partner, so what? Prove me wrong and release the first HBM MXM module this year before nGreedia can get this title as well, because they will... after another year though.
If AMD release a full blown R9 Nano in a laptop, I'll probably get that with 6700k cpu.
It would be very odd if OEM's gave preference to Nvidia in terms of being willing to redesign a whole laptop just to fit a desktop 980 inside, whereas an R9 Nano which is also smaller and has a lower TDP compared to what 980 was advertised for (up to 200W) would be left on the sidelines.
Then again, OEM preference for Intel and Nvidia is rather plainly evident what with the APU's getting stuck with relatively cheap/underwhelming hardware and garbage cooling solutions.
GM204 is 165W, Fiji is 275W but throttled to 175W in Nano. The 200W refers to the card's overbuilt power delivery (more robust than even desktop 980) and the fact that the P870DM's entire GPU cooling solution, which is designed to dissipate the heat of 2x100W MXM cards, is dedicated to it.
Well if APUs actually had some sort of processing power behind them, they'd probably be used more (and I'm not talking about the iGPU).
When AMD's Zen comes out, IF it's good, and if it's introduced in APU formats, we might see better classes of APU notebooks. It's a very difficult thing to argue for when ULV intel CPUs wholeheartedly succeed the desktop's best APUs quite a lot of the time; far less the mobile ones.
I remember earlier APUs bottlenecking even a single 7970M so hard that in Bioshock Infinite, "low" and "high" had the same FPS (41).
AMD has never been a big contender in the mobile area. They do have launched solid products but their portfolio is a mere fraction of nVidias. You might get about 3 GPUs in total, for several years.
I would love to see something like a mobile Nano, but I think at this point it is a waste of money for them. They can easily wait for new fabrication process to make a complete Nano at acceptable TDP for laptops.
As much as I am interested in silly nvidia doing full 980 desktop into laptop, I am perfectly happy with my cool running 980m.
AMD has never been a big player in the mobile space.
For at least the past 10 years, AMD mobile GPUs have always ran hotter, drew more power, and/or performed below the contemporary-generation nVidia equivalent product. They spend less than nVidia on GPU R&D by a factor of 1:3. They spend less than nVidia on driver development by a factor of 1:10. The R9 Nano is the only product we've seen in the past 8 years that emphasizes low power draw.
So forget asking for AMD's response to the GTX 980 in notebooks. When was the last time you saw an AMD mobile GPU *at all*? They just aren't effective at making and selling laptop GPUs.
But later iterations of APU's weren't bottlenecking the dGPU's as far as I could tell - or at least it was minimized.
Regardless, that doesn't excuse OEM's from putting garbage cooling solutions which results in underperforming hardware.
How is the hardware supposed to perform at its own optimum if it cannot be sufficiently cooled?
That's not AMD's problem, that's OEM's problem who kept getting paid by Intel to use their own products with better solutions, whereas then AMD got the short end of the stick (and the same thing kept repeating itself with Kaveri and now Carrizo).
The only decent Carrizo options I've seen were in the EU, but we don't know if those are 15W or 35W.
Other hardware seems decent along with 1080p screens (which is certainly a step up from what is sold in the USA), but we also don't know about cooling, and other things.
OEM's in the USA seem to prefer using 15W Carrizo as opposed to the more powerful 35W ones.
It is ridiculous... but AMD cannot do that much in this area - its not really up to AMD really, and the OEM's are probably getting cash from Intel, so AMD keeps getting short end of the stick.
Regardless of how good Zen will be, it alone probably won't change that.
Look at what AMD did with Carrizo on the same manuf. process.
They cut power consumption by 40% at least and increased iGP performance by over 20% (I think... or more) while also increasing IPC on the CPU side in single-thread by about 5%, while multithreading got a boost of about 10 - 15%.
They actually pulled off something good with an older architecture to keep themselves afloat... and a lot of people ARE interested in AMD APU's, but no one can actually get any information about a decent product out there.
What good though is power efficiency if the processor is so S-L-O-W and so T-H-R-O-T-T-L-E-D? AMD (Always Missing Deliveries) needs kick butt processing power and better GPU performance if they intend to stay in the game. Any less won't help them. Got to be competitive, cheaper, and power efficient.
Not only that but Iam pretty sure nvidia revised the 680m (was gonna be the 675mx) once they found out what they were up against.
Sad to recall really. It was an exciting time for laptops. The pinnacle! When competition existed at the high end and when Alienware made good machines!
Again up to the OEM/ODMs. AMD don't have the cash to say - get this money and make me a good laptop. They supply the parts and the OEM/ODMs decide how to implement them. That's usually how it should be done, unless this -> The grIntel/nGreedia is the other way around, they command everything, throw money for good measure and (most) OEM/ODMs say "YES SIR" and call it a day. They get their cut, so why should they care how the balance of the world turns out in the end? Hence the BGA transition, hence poor AMD setups, hence pretty much everything bad (from costumer point of view) that happens lately.
It wont. You can fit just a few mm larger chip putting it at a slight angle and the HBMs are quite a lot larger than this.
AMD processor being 'slow' is entirely up to the user workload and arguably subjective.
APU's are more than good enough for most people who use computers for general purposes, and their overall HSA capabilities put any Intel processor to shame (most software on the market hasn't caught up yet however).
So, it's not fair to simply compare their cpu capabilities - you have to take into account other things.
For Photoshop and programs that support HSA, AMD showed to be consistently superior - not to mention gaming.
As for throttling... it's mainly there to prevent overheating. Proper cooling solutions would eliminate the problem along with adequate voltage regulation.
Earlier iterations of APU's had issues with throttling, but later ones like Kaveri and Carrizo do not seem to suffer from this problem.
Also, most OEM's pair AMD APU's with slowest possible HDD's and single channel RAM, whereas APU's prefer dual or quad channel RAM. In case of DDR3, best would be using 1800 and 2100 MhZ with very low timings - but OEM's usually pair them with slow RAM and high timings, and also put lousy screens on them - even though the iGP on APU's were shown they can run several modern games on medium details in 1080p (which is where they started to show better performance compared to Intel vs lower resolutions that are more CPU dependent).
Minimized, maybe, but even their best desktop APU can bottleneck a 780M, far less a decent 970M or 980M. They really have no place in the power market right now, especially with laptops.
Yes, I wholeheartedly agree. But OEMs are making them for the cheapest notebooks, and that means cutbacks everywhere. They're hotter than the intel ULV chips they're competing with (as well as weaker generally), so why would an OEM pick up a machine that needs a thicker/heavier machine that performs worse than the ULV chips? Especially because (I'm using your later point which I'll expand on more) for these kinds of machines, the intel iGPU is more than good enough for general purposes? Especially at the lower tiers (sub $700) where OEMs want to save the most they can on internals while having the machine look/feel attractive. Now I'm not saying that making a decent $700-$800 medium-thickness sort of basic-gaming media laptop is a bad idea, but for the same reason the whole BGA craze is going strong right now, this won't happen: nobody wants to buy a machine thicker than they THINK will get the job done. And this is the biggest problem: it's not "thicker than is necessary", but "thicker than they THINK is necessary". This is why the Razer Blade and the Aorus laptops are praised by the mass market who don't care if something overheats or throttles as long as their game doesn't "feel worse".
As much as I both know Intel has done it in the past (and paid for it by lawsuit) AND that I wouldn't think to put it past them, I don't think it's happening. I think the APU's iGPU is of enough less importance since intel's iGPUs have stopped being COMPLETE garbage (still garbage. Just not 100% garbage. =D) and the intel CPUs are that much better computationally and temperature-wise that it's simply more expensive to put in an APU in a system. Even if the companies are only losing $0.50 per unit created by making them with APUs instead of ULV CPUs, they still want to save that.
I won't argue with you there. But AMD's main issue is that their APUs need to have a massive IPC bump keeping (or improving) the efficiency they have now. The sad truth is that while intel is slowly milking the market, they are reliable, and it's not an argue-able point that each new generation of CPUs is better in PERFORMANCE than the last. Overclockability is an entirely different story. But get them running at the same speeds? The new chips will plainly and simply be better. AMD on the other hand has a lot to prove. A ridiculous amount. Their APU design in itself isn't bad in theory. I'd like them to properly improve the tech. But I'll be damned if I give them a pass on a low IPC generation after generation because their iGPU is better, when for basic purposes, the competition does its job.
Comparing what's currently available on the market, they are slow. It's not subjective, it's a fact. The difference can be more apparent depending on the workloads, but it doesn't make them faster or slower.
I expanded a lot on this above, but I'll talk about the photoshop comment. Buying a very cheap laptop with an APU in it to use somewhat CPU heavy tasks like photoshop is just... counter-productive. Yes, some people might do it, but a better idea would be to buy one of the quadcore i7 chips on the whole. Working around with previews in those programs can put a decent strain on the CPU too, not just outputting the files.
Hand me a Carrizo and we'll see if it does not suffer from the problem.
Again: this is a result of the tier of notebooks. If the APUs had better CPU power and were good enough to be like those machines that come with the i5/i7 M/MQ/H/HQ chips and no dGPU, then we wouldn't be NEARLY in this problem. But even with those better machines, good RAM is scarce because people believe RAM is RAM and timings mean nothing. I've even spent hours trying to explain to VARIOUS PEOPLE that going from 1600MHz 11-11-11-27 RAM to 2133MHz 11-11-11-27 RAM is in EVERY REGARD an upgrade, when they keep pointing to benchmarks showing 1600MHz (8-8-8-20 or better) and 2133MHz (11-11-11-27 or worse) RAM have little functional differences, and failing to realize that high-latency "average speed" RAM is still "bad". Try telling less technically adept people that they need to "pay extra" for "non-bottom-tier" RAM and that there is almost no evidence online to show that the better RAM will help them in some way. See how many people adopt it. Basically? APUs were designed to be in midrange, dGPU-less media/creation/etc machines that can do some light gaming. Their IPC is holding them back from keeping this slot. Just look at how much AMD's FX chips hold back even last gen cards in gaming, far less what they do when TDP limited and in a laptop.
And I want to make one thing very clear: putting a CPU in a scenario where a GPU bottleneck is forced much more easily does not give the CPU a free pass. You said that 1080p gaming is more where the APU shines because CPUs don't matter as much. That is false. What happens is that the iGPUs are too weak at those resolutions and get limited quickly, before the CPU does. It doesn't mean the CPU is better in any way. It just means you've offset the point where it shows its crippling feature. But again, pair it with a dGPU and that plan goes down the drain real fast. I have to file the 1080p screen etc to show "how bad the intel iGPU is" as a "cherry picking" scenario. It's the same reason I roll my eyes at benchmark compilations meant to prove "AMD CPUs are fine for gaming" and they pick a bunch of games that either couldn't care less if you were on a core 2 duo (like Tomb Raider 2013), uses basically every inch of CPU you toss at it (like BF4; granting a rare instance where the full 8-core chips actually mean something), or is in a clear GPU bottleneck (like Sleeping Dogs with Extreme AA aka 4x SSAA enabled and a single R9 290X).
If APU is so much better on processing power, then why is Intel Core i7 being used for advanced PhotoShop users and why do the high end Gaming and Workstation laptops have Core i7, it's simple. AMD is not there yet. They are behind by years. They won't get installed into higher end machines without kick butt processing power and better GPU to back it up. Until AMD sinks effort and money into this, they will be left behind. AMD had it when Athlon was released, they've not hit that height ever since. And it will take a major shift of thinking and architecture to get AMD there, which is major effort and expense. I want to see AMD succeed. But AMD has to want to do it and have the resources to do it. I like competitive playing fields.
Yes, the TDP is an
assumption made about the cooling capacity of the platform the cpu is installed in. It has no relation to actual circumstances. You could provide unlimited juice plus zero-Kelvin temperatures and it's
still down-throttle when TDP is reached.
Actual thermal throttle is different, of course, but that shouldn't kick in before 95-97°C.
It's not. Not even close. He's talking about a technology called HSA, which beats intel's hardware acceleration by a mile for programs like Photoshop that utilize it.
It's like saying that the intel iGPU is more efficient for rendering due to quicksync acceleration than the CPU (which is correct). But it's not got anything to do with comparing CPU power.
Yes, and this in itself is a problem for the higher end chips, as you know =D. No point having a $600 CPU that refuses to use more than 47W under load. Welcome to the BGA-hate train <3
Can you explain a bit more about how this works? In Photoshop CC speed tests and benchmarks I've seen, the eight-core AMD processors seem to perform poorly compared to quad core i7s with similar clock speed.
Well ... not $600, or even half that (ES + discount), but yes ... it seriously spoils the fun.
Anyone ever compared the values of those tiny ICs on the cpu itself? And what happened to good 'ol pin modding? Hmm ... Intel datasheets aren't as forthcoming with Haswell, it seems.
I've also wondered what AMD is looking at with HBM for mobile parts. Given the smaller overall size, it seems like something they must be looking at. NVIDIA did pull a surprise with the laptop 980, which suggests that it shouldn't be too difficult for AMD to pull off something similar and Fury Nano based. triturbo may have hit on the reason with the MXM specs in post 3. But nonetheless, I'm sure AMD's looking at how they can introduce HBM into laptops... if not this year, I'd expect to at least see it with HBM 2.0 and 14nm chips next year, and they'd have good reason to make sure they get it out in a timely fashion while they enjoy HBM 2.0 priority.
The link didn't go to any particular benchmark, but you're probably looking at Vishera benchmarks (ex. FX-8350), which don't have HSA. AMD's newer APUs do support HSA, so if Photoshop benefits from it, you'd see this in the APUs (mobile and desktop) pulling more than their expected weight based on purely CPU benchmarks, but wouldn't see it in the desktop FX series at this point.
Do you mean like the Kaveri A10 quad cores, or something else - which AMD models get this boost for Photoshop? And is it only for one specific kind of task, or general usage scenario?
I beg to differ. 5870M was better than the 480M, 6990M was equal to the 580M and beat it in some areas (and vice versa), but both cards were horrendously unreliable. The 7970M came out long before the 680M and still trades blows with it to this day, and absolutely demolishes it in professional applications due to nGreedia neutering the 680M's professional cores. The only time AMD became irrelevant is when nVidia introduced the 780M and AMD began its 3 year long rebrand run.
That is not the point. It is not about performance competition. You basically named like 60% of the entire portfolio of amd gpus in the last several years.
The point was that nvidia release a lot more gpus for different levels of perfoance thus being a big competitor compared amd.
At this point, I wouldn't even care if what they introduced wasn't even modular (just until the next gen of course). Just something - ANYTHING - that can compete with a 980M. Hell, make a mobile Nano that's soldered for all I care. The big sellers are laptops like the P650SG anyway.
Yep, I mentioned this a few years ago when 7970M was still top dog. The segmentation in Nvidia's mobile product stack means it has a GPU to hit every price and performance point. AMD had only 7970M at the top, and next down was 7870M which was less than half as fast. Such a huge performance and price gap which Nvidia filled with like 10 different products and had no competition. This equals a wider range of systems offering Nvidia GPUs, and more profit for Nvidia as the mainstream market that buys laptops with midrange GPUs is much bigger than the enthusiast market that spends top dollar for a flagship GPU like the 7970M. Nvidia had no competition in the most lucrative segment of the market.
And it's not like AMD didn't have anything to offer there, they could have easily spun Pitcairn Pro and Bonaire into some very competitive mobile GPUs to slot between Cape Verde and Pitcairn XT. But stupid is as stupid does and they played the apathy card with the mobile market. By the time they woke up, Nvidia had completely taken over.
Isn't it funny that both nVidia and AMD have this issue in the desktop market right now?
Look at the price (and performance) jumps between the 960, 970, 980 and 980Ti. The only cards that make sense for the dollar are the broken 970 and the 980Ti. They could have literally cleaned stock with a 960Ti at the ~$240 price range based off the 970M and made the 960 a $180 card (or simply called it a 950Ti and made the 970M into a 960 with the same prices) and they'd have a great stopgap to the 970 range. But noooo.
And on AMD's side it's not even great either: the R9 390 is fantastic for its price, but it needs such a large PSU compared to a 970 that some users actually rather go with cheaper/smaller machines, which is the whole issue I had with AMD in the first place: your whole line can't be hot and power hungry. At least with the R9 380X coming out they'll have a great stopgap card; the differences between the 380/960 and the 390/970 are so huge.
The 390X kinda makes sense as well.~$400 for a card that goes head to head with a 980. AMD also has an advantage with the Fury, since Nvidia has nothing in the $500-$600 price range right now.
Honestly, it hurts me to see how low AMD needs to cut their prices to do well, but I can't imagine nVidia cards being bought out greatly (even the 980Ti) if AMD's price line was something like this:
Fury X: $550
Fury: $480
R9 390X: $390
R9 390: $280
R9 380X: $230
R9 380: $170
But of course who knows how much that'd make them bleed cash. But hey, they'd be getting a lot of sales I think. Even people who hunt EVGA B-stock for nVidia cards would have to stop and think twice about those cards.
Oh I know. They just don't like to take our credit cards and stuff. Just like with XoticPC; they refuse to take credit cards that aren't USA and require a Bank Wire Transfer.
It's much more difficult than you know to get some of these websites to work =D. We already use Amazon and various sites like that with Skyboxes already. But some sites are just harder than others.
Also, the first website you listed is EXTREMELY expensive compared to what we get with some of our skyboxes down here =D
Problems? See this thread at archive.org.
![[IMG]](images/storyImages/large.gif)
D2 Ultima likes this.
.
