AMD - Piledriver chitchat

[dfour]

Next years no damn good they need it out this year at the latest.

[ik9000]

Next years no damn good they need it out this year at the latest.

Quite. And that's if it materialises at all. There are conflicting accounts of what to expect as previous posts link to. And then there's the inevitable over-hyping and under performing that sadly seems to go with most AMD CPU/APU releases of late. Having waited and waited for the hailed bulldozer only to see it be a disappointment I can't be bothered waiting another 2 years this time round. I'll either buy an FX8350 or i7 in the next few months. Still haven't decided yet whether I want to flick two fingers at Intel stripping off virtualisation and the like (or else locking the chip down) for no good reason, or plump to support the underdog with their unlocked chips, and all features, but which are behind intel on power consumption and single thread performance. The motherboard availability (or lack thereof) for more recent features put me off going AMD. They need to do a newer chipset to improve on the AM3+ 990FX.

[DanceswithUnix]

Quite. And that's if it materialises at all. There are conflicting accounts of what to expect as previous posts link to. And then there's the inevitable over-hyping and under performing that sadly seems to go with most AMD CPU/APU releases of late. Having waited and waited for the hailed bulldozer only to see it be a disappointment I can't be bothered waiting another 2 years this time round. I'll either buy an FX8350 or i7 in the next few months. Still haven't decided yet whether I want to flick two fingers at Intel stripping off virtualisation and the like (or else locking the chip down) for no good reason, or plump to support the underdog with their unlocked chips, and all features, but which are behind intel on power consumption and single thread performance. The motherboard availability (or lack thereof) for more recent features put me off going AMD. They need to do a newer chipset to improve on the AM3+ 990FX.

Only thing I would look forward to in the next generation is the process shrink. 32nm to 28nm bought us pretty much nothing on the CPU, but I would hope that a shrink to 20nm would bring us not just more cores but perhaps some L3 cache too. At 6 updated cores it would start to make the FX line questionable.

[ik9000]

Only thing I would look forward to in the next generation is the process shrink. 32nm to 28nm bought us pretty much nothing on the CPU, but I would hope that a shrink to 20nm would bring us not just more cores but perhaps some L3 cache too. At 6 updated cores it would start to make the FX line questionable.

but it's not worth waiting to end 2015 for.

[Noxvayl]

Well they have made architecture improvements in their APU's while their top end CPU's have not been updated so you can expect those to be carried over to the new top end CPU's. Perhaps they are looking to have the high end get larger jumps between generations and do the iterative changes in APU's instead.

[ik9000]

Well they have made architecture improvements in their APU's while their top end CPU's have not been updated so you can expect those to be carried over to the new top end CPU's. Perhaps they are looking to have the high end get larger jumps between generations and do the iterative changes in APU's instead.

But they won't be out for ages. Even if the 2015 rumour is true, it won't therefore actually be out therefore until mid 2016 at the latest. And if that's being harsh then IIRC Bulldozer - delayed. Piledriver - delayed. Kaveri - delayed. Mantle, still to be properly released...

[DanceswithUnix]

but it's not worth waiting to end 2015 for.

That's why I got an FX8350 9 months ago. For me I think it was a really good choice, and if software really is getting more threaded then I think I am good for another year or two.

Well they have made architecture improvements in their APU's while their top end CPU's have not been updated so you can expect those to be carried over to the new top end CPU's. Perhaps they are looking to have the high end get larger jumps between generations and do the iterative changes in APU's instead.

I think they just ran out of resource and had to stop development of something, and AM3+ is what got the chop. When DDR4 comes out, I wonder if they will bother with an AM4 platform, but if they do abandon the old CPU platform then I hope they at least do a desktop variant of the FMx. Not like the FM2 Athlon where they fuse some stuff off, a proper silicon variant with hardly any graphics shaders but plenty of cores and L3 cache.

[watercooled]

Because no-one besides AMD over-hypes releases, or delays them?

Kaveri is price, performance, power and die-size (comparing cores; of couse Kaveri has a larger and far more powerful GPU) competitive with i3, which the market it's aimed at. Bear in mind, again, that an Intel core is roughly the same size, on the same node, as an AMD module. And I wouldn't say the 28nm node brought nothing, although it's of course hard to precisely quantify it because we're also dealing with a modified uarch, but especially at lower clocks i.e. <65W desktop parts or mobile TDPs, Kaveri brings some significant efficiency improvements over Richland.

[Noxvayl]

But they won't be out for ages. Even if the 2015 rumour is true, it won't therefore actually be out therefore until mid 2016 at the latest. And if that's being harsh then IIRC Bulldozer - delayed. Piledriver - delayed. Kaveri - delayed. Mantle, still to be properly released...

I agree that the wait is crazy. 3 years between high end CPU releases is pushing it, I don't think there are too many people looking to hold onto their processors for 3-5 years so it makes little sense to me but it does seem to be what is happening unfortunately. I do expect that when the release comes it will be a significant performance jump though, with all the iterations done for their APU's they better have learnt enough to make a chip with a significant performance boost over the FX8350.

That's why I got an FX8350 9 months ago. For me I think it was a really good choice, and if software really is getting more threaded then I think I am good for another year or two.



I think they just ran out of resource and had to stop development of something, and AM3+ is what got the chop. When DDR4 comes out, I wonder if they will bother with an AM4 platform, but if they do abandon the old CPU platform then I hope they at least do a desktop variant of the FMx. Not like the FM2 Athlon where they fuse some stuff off, a proper silicon variant with hardly any graphics shaders but plenty of cores and L3 cache.

That is a good way to look at it. I hope things turn around enough for development to be more frequent at the top end going forward.

I've been having issues with my i5 2500K and MSi Z77-G43 Gaming motherboard. Isolated the memory running at 2133MHz as the problem but warranty won't cover that so I've decided to put some money into a fully featured motherboard and the best value for the features I'd like was for the Asus Crosshair V Formula-Z supporting the FX8350, so I've gone for that. I expect games going forward to support the processor better but at the moment Excel and AutoCAD, which I use a lot of, run better on the FX8350 than Intel's processors and it is cheaper so I'm all aboard the AMD thread train.

[CAT-THE-FIFTH]

It seems a Richland based Brix has been released:

http://www.ebuyer.com/625808-gigabyt...e-gb-bxa8-5545

Its cheaper than any of the Core i3 based Brix systems too.

It uses the A8-5545M.

[CAT-THE-FIFTH]

Wow,another rubbish article from TR:

http://techreport.com/review/26082/t...system-guide/2

Some of you might still have reservations about buying a dual-core, quad-thread processor when the latest consoles feature eight-core chips. Won't the new breed of cross-platform games need just as many cores on the PC? Well, no. The processors inside the PlayStation 4 and Xbox One are based on AMD's lightweight Jaguar architecture, which is far slower clock-for-clock than Haswell. Our Core i3 could easily do the same work with its two cores and four threads. Also, everything we know about game programming tells us that, at least on the PC, single-threaded performance remains very important. In the words of Jurjen Katsman, one of the guys behind the PC versions of Thief and Deus Ex: Human Revolution, most games "flatten off at one core."

In short, the Core i3-4130 is a fine choice for a budget gaming build. We'd certainly recommend it over similarly priced alternatives from AMD like the FX-6300, which is saddled with poor single-threaded performance, high power consumption, and an outdated platform. AMD's Kaveri APU might make a decent alternative, but alas, no variants of it are currently available at this price.

So why doesn't TR quote companies like DICE which have pushed threaded games better??

It is a distinct lack of insight,especially since all the main engines are designed to thread better.

Its like the E8400 against Q6600 arguments. The Q6600 won convincingly.

If that is the case TR,why don't you specify a Celeron dual core??

Why do you even put Core i5 and Core i7 CPUs in your top end rigs?

Or so multi-threaded performance is important as long as it is Intel??

I like how they ignored the FX6300/FX6350 being faster than the IB Core i3 in many games:

http://techreport.com/review/24954/a...pus-reviewed/6









Moreover,in C3 they did not test "Welcome to the Jungle" which is the most CPU intensive part,where a Core i3 does much worse.

Where is this advantage TR??

Yet,even months later still recommended the Core i3 3220:

http://techreport.com/review/25250/t...system-guide/2

Whats the bet if they test newer games and the FX series is ahead of the Core i3 CPUs,they will start pushing the "but you can upgrade the Intel CPU" mantra.

TR is massively loosing credibility at all levels,especially after that GTX750TI review.

It seems the author has an agenda,when even his own reviews don't say the same thing. It means he pre-judges issues,and is not objective as a result.

It looks like Intel and Nvidia PR works very well on them.

Their inability to read their own charts properly,or not realise inconsistencies is starting to verge on Bit-tech levels.

[mikeo01]

I mean I'm not the best at analysing articles but they just seem to be punching AMD for no reason here

So essentially they are saying "Don't buy AMD everyone even that cheap I3 can beat anything they have to offer". That's how I am interpreting it.


Bleugh

[CAT-THE-FIFTH]

Either they are Intel fans deep at heart,Intel and Nvidia PR is more enticing or they simply make more dosh from recommending them.
Its telling when the FX6300/FX6350 were beating the Core i3 3220,and they still recommended the IB Core i3.

There is real need for a fully independent review site!!

[mikeo01]

Probably a mix of all three to be honest. The FX-6300 seems better value for money anyway which they didn't say.

I mean for a reviewer to go against their own benchmark results it says it all really. It's always the same, reviewers shrug when something of AMD's actually performs decent, as if it's a one off result and not worth considering.

AMD need to step up to this and start taking them seriously as it does hurt AMD's business in the market. Unsuspecting buyers trust reviews.

[watercooled]

Thing is, the are *very* few games I am aware of where an i3 performs more than on-par with similarly priced AMD CPUs (ignoring the cherry-picked and pointless 640x480-type tests of course). However there are plenty of games (i.e. better threaded ones) where a dual core i3 falls *significantly* behind quad+ cores, and will have 4 threads running flat out, especially as you say in CPU intensive parts, or playing online, etc.

Also, if someone wants to do any processing on their system, like video encoding etc, an FX can be well over twice as fast (in fact FX is often on-par or slightly better than even i7s here).

So, you get a CPU which is on-par in some games, lags behind significantly in others, and falls well behind in most well-threaded desktop processing.

Unfortunately an awful lot of reviews these days seem to be all about plain cherry-picking results to meet a motive. Like you say they've seemingly completely ignored even their own testing results.

Edit: Yeah it's funny how AMD are downplayed even when they are conclusively ahead e.g. IGP. It was suddenly important to certain websites when GT3e was released, but now AMD are back well in the lead, and for far less money, and good power efficiency, it's not important any more!!

Edit2: Actually aside from a few deprecated benchmarks like SuperPi, the difference in single-threaded performance really isn't all that different. Certainly nothing compared to the huge chasm in multithreaded performance.

[watercooled]

Speaking of bittech, what were they smoking when they came up with their Handbrake benchmark? http://www.bit-tech.net/hardware/201...swell-review/3

A 6 core, 12-thread i7 is really less than twice as fast as an i3? Really? What were they transcoding, a dot?