The AMD Execution Thread [2007 - 2017]

Status
Not open for further replies.
At worst, Zen will be as solid in being number two as Bulldozer is right now.

I mean, no one is really expecting VIA's 65nm Nano cores to take over AMD's offerings, so...
 
Last edited by a moderator:
At worst, Zen will as solid in being number two as Bulldozer is right now.

There is little justice for AMD as a company. :(

Even if they unleash the power with something super fast, there will be the competition with its dirty tricks stealing all the fame.

Remember the Athlon 64 times when it was faster than Pentiums and still most o0f the time consumers bought pentiums?!
 
There is little justice for AMD as a company. :(

Even if they unleash the power with something super fast, there will be the competition with its dirty tricks stealing all the fame.

Intel didn't force AMD to design Bulldozer. They made that choice themselves, they screwed up. Intel don't even have to bother with their anti-competitive practices any more.
 
Intel didn't force AMD to design Bulldozer. They made that choice themselves, they screwed up. Intel don't even have to bother with their anti-competitive practices any more.


Are you 100% sure of that? There are things that make me cringe.
Take all the AMD APU/CPU reviews. Here we have AMD developing an API that allows a CPU with lower performance (like the ones from AMD) to perform better in games. Mantle should also be a good indicator of the performance that DX12 games would get on lower performing CPUs, and not only the ones from AMD.
Yet, go to Anandtech and they simply refuse to show a single result of an AMD APU and a discrete graphics card using Mantle.

And what about the actual Mantle reviews? Well, let's show how useless Mantle is by pairing it with the most expensive of the CPUs that AMD themselves claim will get the lowest performance boost.
I can't find any major website that tests AMD's recent APUs with Mantle, nor could I find a Mantle review that didn't use a >$400 Intel Core i7 or better.

Coincidence? Maybe.
 
I can't find any major website that tests AMD's recent APUs with Mantle, nor could I find a Mantle review that didn't use a >$400 Intel Core i7 or better.

Coincidence? Maybe.

Incompetence, I'd say. There's a long history amongst tech sites of missing the point in this manner - eg. let's test gaming performance of Celeron & Pentium using top-end graphics card and 1kW PSU, or lets build an HTPC using a top-end i7 and GPU. Very few engage brain and put together systems that reflect where the components of interest are aimed in the market.
 
Incompetence, I'd say. There's a long history amongst tech sites of missing the point in this manner

On Anandtech? There's really little hope for tech journalism if their flagship site is simply ignorant at this level..
 
On Anandtech? There's really little hope for tech journalism if their flagship site is simply ignorant at this level..

Yes, on Anandtech. They have got a little bit better recently, especially on their HTPC stuff, but many a time I've banged my head against my monitor trying to figure out a decent low- to mid-range non-gaming build, or low-power or low-noise build based on reviews at AT, Toms, etc.
 
On Anandtech? There's really little hope for tech journalism if their flagship site is simply ignorant at this level..
Flagship? Probably depends on how you define tech journalism...On a separate note, UniversalTruth, I believe I've already told you not to bring the fanboy-ish ditch-radio tirades in, haven't I?
 
On a separate note, UniversalTruth, I believe I've already told you not to bring the fanboy-ish ditch-radio tirades in, haven't I?

Yes, but I do not think I violate anything at the moment. It is my personal opinion based on my observations on the surrounding world. If I think it is like this, there is a high probability it is indeed ;)

339rdiq.jpg


Anyways, look at how AMD A10-7870K BEATS i3 + 740 GT.

The question is how many people do know this and would buy the A10?
 
Yes, on Anandtech. They have got a little bit better recently, especially on their HTPC stuff, but many a time I've banged my head against my monitor trying to figure out a decent low- to mid-range non-gaming build, or low-power or low-noise build based on reviews at AT, Toms, etc.
That's why you use HTPC dedicated sites. One of the best sources for that is the Kodi forums.
 
That's why you use HTPC dedicated sites. One of the best sources for that is the Kodi forums.

The specialist sites only go so far. Kodi don't do HTPC hardware reviews - the forums are a bunch of individuals with their own requirements and standards, and generally speaking they're just rating the kit they have bought themselves (I hang around there a lot!). Most people there can't afford to buy half-a-dozen CPUs to compare against each other. Likewise SPCR is great for some elements of the low-noise, low-power domain, but only addresses part of the question.
 
Anyways, look at how AMD A10-7870K BEATS i3 + 740 GT.

The question is how many people do know this and would buy the A10?
I wish you'd post a graph other than this marketing slide, besides the obvious cherrypicking of results, the bars don't start from 0, so the difference is probably exaggerated.
 
What system RAM was used in that marketing slide?
Unlike the A10, the i3 doesn't require faster and more expensive RAM to perform well.
When one looks at whole system cost and upgradability then the A10 just doesn't make sense. A cheaper CPU + cheaper RAM + GDDR5 videocard is better.
 
I wish you'd post a graph other than this marketing slide, besides the obvious cherrypicking of results, the bars don't start from 0, so the difference is probably exaggerated.
I don't disagree with the rest of your post, but a quick pixel count estimate shows that the bars start from 0 or very close to 0 (they are all about 2.92-2.94 pixels/FPS).
 
Yet, go to Anandtech and they simply refuse to show a single result of an AMD APU and a discrete graphics card using Mantle.
I would note that none of those games feature Mantle. And that's because the only games with Mantle right now are Frostbite engine games (which have their issues, especially for CPU testing), Thief, and Civ:BE. Given our goal of testing a wide range of genres in our suite, the age of various games, and what we were trying to test, Thief and Civ:BE were not the kind of games we were looking for. This is not in any way a snub to Mantle (go take a look at our GPU reviews where we do include these games), but rather it reflects the fact that Mantle is currently in very few games, and none of those align with our CPU testing needs.

And what about the actual Mantle reviews? Well, let's show how useless Mantle is by pairing it with the most expensive of the CPUs that AMD themselves claim will get the lowest performance boost.
I can't find any major website that tests AMD's recent APUs with Mantle, nor could I find a Mantle review that didn't use a >$400 Intel Core i7 or better.
I would also note that we did in fact test BF4 at various CPU speeds using our standard GPU testbench i7 clocked at different speeds to simulate other results, precisely to see how lower performance processors would be impacted.

That said, the greatest gains for these early Mantle games in terms of average FPS is going to be when you have a combination of a very powerful GPU with a very weak CPU, so that the bottleneck is draw calls coming off of the CPU. This is an unrealistic scenario; if you're buying a high-end system you're going to want a fast processor to go with it. Otherwise such an unbalanced system will work well in a few Mantle games, but fall very short everywhere else.

Anyhow, we have done quite a bit of coverage of low-level APIs over the last year or so. So if you haven't already, I'd encourage you to check our our DX12 coverage, our Civilization coverage, and of course everything else we've written on Mantle.:)
 
Also, the benchmarking situation will become way better when DX12 games arrive. Major game engines will support DX12. DX12/Mantle advantages are similar.
Hence, there will be far more chance for AMD APUs to shine.
Though, since this is the doom thread, let me add a gloomy note: due to AMD's lack of resources, Intel APUs and Nvidia GPUs will actually be better supported, even though they are less CPU-limited (Haswell and Fermi are getting DX12 support while Trinity/Richland will stay on DX11).
 
I wish you'd post a graph other than this marketing slide, besides the obvious cherrypicking of results, the bars don't start from 0, so the difference is probably exaggerated.

Actually, before making accusations, how about actually looking at the graph. It's obvious it starts from 0 if you compare the relative positions of say DOTA 2 at 49 FPS to LoL at 89 FPS. Or any of the other bars.

Now, that may or may not be representative of non-Esports online games, but the graph isn't exaggerating any differences in performance.

And if review sites would more extensively test budget systems, then we wouldn't have to rely on marketing slides.

But considering that it's mostly enthusiasts or performance level gamers that go to review sites, there isn't much incentive for them to do reviews on budget level gear. So it's hard to blame tech sites for catering to the audience that is most likely to go to their web sites.

Regards,
SB
 
What system RAM was used in that marketing slide?
Unlike the A10, the i3 doesn't require faster and more expensive RAM to perform well.

A10-7870K + 8 GB DDR3-2400 vs i3 4360 + GT 740 1 GB DDR3 + 8 GB DDR3-1600.

http://www.techpowerup.com/212948/amd-announces-new-a-series-desktop-apus.html

1. Testing by AMD Performance Labs using: AMD A10-7870K with AMD Radeon R7 Graphics, 2x4GB DDR3-2400, 256 SSD, Windows 8.1 64 bit, Driver 14.502. Core i3 4360 with Nvidia GT 740 Graphics 1GB DDR3, 2x4GB DDR3-1600, 256 SSD, Windows 8.1 64 bit, Driver 4156. Games tested: CounterStrike:GO @ 1080P Max settings (AMD platform scored 62fps vs 46fps), DOTA2 @ 1080P Max settings (AMD platform scored 49fps vs 37fps), League of Legends @1080p Max settings (AMD platform scored 89fps vs 60fps), StarCraft II run at max settings at 1080p (AMD platform scored 35fps vs 28fps). GV-8

When one looks at whole system cost and upgradability then the A10 just doesn't make sense. A cheaper CPU + cheaper RAM + GDDR5 videocard is better.

  • The AMD A10-7870K APU features 12 compute cores (4 CPU + 8 GPU)*, up to 4.1 GHz CPU clock and 95W TDP for exceptional performance on today's modern workloads, and up to 5 percent faster system performance versus a competitive processor.5
5. Based on PCMark Home testing using an AMD A10-7870K with 2x8 GB DDR3-2400, 256GB SSD, Windows 8.1, beta 15.10 drivers which scored 3457 in PCMark 8 Home v2. Intel i3 4370 with 2x8 DDR3-2133, 256GB SSD, Windows 8.1 Driver 4156 which scored 3304 in PCMark 8 Home v2. Core i3 4370 pricing of $151US as of May 7, 2015 at pricewatch.com vs $130US for A10-7870K estimated e-tail pricing as of May 28, 2015. GV-1

You can read the press release article.

I do NOT understand your thought about the system cost. This is exactly what the review by AMD Labs was done around.
 
Status
Not open for further replies.
Back
Top