Jump to content
SceptreCore

AMD conFusion? Forget the hype!

Recommended Posts

OMG. Retarded "reviewers" thinking that superpi/hyperpi is a CPU performance metric, when all it is is x87 instruction optimising performance. Something Intel made superfluous by introducing SSE instructions.

 

People who care about superpi results... Don't know whats going on.

Share this post


Link to post
Share on other sites

Ok so after reading a few in depth reviews I really dont see what everyone is making a big deal about. This got hyped to hell my markerting and a lot of people fell for it. Everyone was sitting here hoping for the next athlon x2. This isnt it but its not "that" far behind and if pricing is anything like the current line up this could be competitive.

 

 

For the current guys with 990 boards Its not that bad an option if your looking to upgrade.

I was thinking the exact same thing. When trudging through that shit storm in the earlier pages... I was thinking... Did everyone read different reviews to me?

 

It just manages to best the 2500k in some tasks and falls short in others. So it's not fast in single threaded tests. Oh no. Who here is running a single core in their rig? And as proven... It can roll with the best of Intel for gaming. Which... Be honest, that's what most of us do. We dont really care how long it takes to convert this, and compress that. Clearly orochi didn't bring it for bulldozer. And seems like it was rushed to just get something new out the door. As journos were commenting about the odd cache latencies, shows it didn't really get the testing time it needed. Or maybe these are process issues or something else that can be ironed out.

 

I confess I find it dissatisfying to see so many chucking big sads, and letting their senses escape them.

 

I myself am still quite content with BDs performance, except for the power consumption. Which for someone as familiar with the architecture as i am... Shouldnt be happening. We all saw how llano can one up SB for power efficiency. So I'll keep monitoring the situation, and see if it improves. I can't buy it till then as we West Aussies have to be more power concious.

 

 

+10 mate I was just typing this and then I refreshed the page and you had posted it. Once we see these in peoples PC's and not just the 1st release reviews we will be able to get an idea about power usage which is the big thing for me and a lot of people no doubt

 

OMG. Retarded "reviewers" thinking that superpi/hyperpi is a CPU performance metric, when all it is is x87 instruction optimising performance. Something Intel made superfluous by introducing SSE instructions.

 

People who care about superpi results... Don't know whats going on.

 

Yeah but its a number man and smaller or bigger numbers mean better performance daaaaaaaar. my puter gets 8 seconds yours gets 11 seconds your puter is bad im pro.

Share this post


Link to post
Share on other sites
What's annoying me is all the varied results.

 

It seems when not overclocked... It stays within its specified envelope. So I probably will buy it after all.

 

Yey.

Share this post


Link to post
Share on other sites

Maybe for someone like me wanting to upgrade to something, just getting the PII X6 with my current MB would be best for the price. It definely is not worth it for anyone on a AM3 to get a new AM3+ MB and BD. But really I would see little gain going that route since I kinda have a PII X4 (unlocked) even at a low 3.5ghz.

What about someone with AM3 Phenom II in an AM2+ board?

 

Well wouldnt you need new RAM too then, so Id say no to upgrade if you have a good AM3 CPU.

Share this post


Link to post
Share on other sites

OMG. Retarded "reviewers" thinking that superpi/hyperpi is a CPU performance metric, when all it is is x87 instruction optimising performance. Something Intel made superfluous by introducing SSE instructions.

 

People who care about superpi results... Don't know whats going on.

Yeah but its a number man and smaller or bigger numbers mean better performance daaaaaaaar. my puter gets 8 seconds yours gets 11 seconds your puter is bad im pro.

 

Oh no. Not 80 bit precision instruction. I guess I'll just have to be content wick pwning all Intel CPUs in 256bit AVX, FMA4, and, XOP optimized benches and apps. :(

 

Damn your proness sir. Damn it to hell.

Share this post


Link to post
Share on other sites

Can't see myself picking one of these up.. as most have said, was hoping for a better effort. It can't match 9-10 month old architecture from intel, and cant even match its old architecture sometimes.. seems a bit too hit and miss in performance.

 

so.. now to decide on grabbing a sandy, or waiting for an ivy..

Edited by p0is(+)n

Share this post


Link to post
Share on other sites

Sceptre, it's a far stretch to say it matches Sandy in gaming, it clearly doesn't, especially when an overclocked sandy is likely to run cooler than even a stock bulldozer..

Share this post


Link to post
Share on other sites

Sceptre, it's a far stretch to say it matches Sandy in gaming, it clearly doesn't, especially when an overclocked sandy is likely to run cooler than even a stock bulldozer..

At close to 5.0ghz reviews showed over 450w consumption.. thats pretty hefty too.

 

Our end result on air thus is 4600 MHz on all eight cores. On proper liquid cooling 5 GHz should be a viable target as well. The power draw however is intense as at this stage we consume roughly 430 Watts by stressing all eight CPU cores.

Edited by p0is(+)n

Share this post


Link to post
Share on other sites

Sceptre, it's a far stretch to say it matches Sandy in gaming, it clearly doesn't, especially when an overclocked sandy is likely to run cooler than even a stock bulldozer..

How is almost identical fps averages in high resolution tests.... with sometimes greater from the BD chip.... a far stretch?

 

Let's not bring overclocking into it shall we. Because if we do that... then it's probably more recommendable to tell people to just buy the 1100T and overclock that and save cash. The Intel fanbois have previously stated these objections. So to this Fanbois pride as with regards BD's power consumption when overclocked. We'll just stick with stock comparisons. :)

 

You cannot seriously disagree with me that BD is high end gaming performer

Share this post


Link to post
Share on other sites

Nobody is saying that it sucks, just that its been hyped up to the heavens and its barely as good as whats already out.

Share this post


Link to post
Share on other sites

exactly they may aswell bottlenck it a bit more then even a p4 could perform the same in thos tests to becuse they are benching the gpu! not the cpu its just misleading bs

Share this post


Link to post
Share on other sites

http://www.anandtech.com/show/4955/the-bul...x8150-tested/11

 

AMD also shared with us that Windows 7 isn't really all that optimized for Bulldozer. Given AMD's unique multi-core module architecture, the OS scheduler needs to know when to place threads on a single module (with shared caches) vs. on separate modules with dedicated caches. Windows 7's scheduler isn't aware of Bulldozer's architecture and as a result sort of places threads wherever it sees fit, regardless of optimal placement. Windows 8 is expected to correct this, however given the short lead time on Bulldozer reviews we weren't able to do much experimenting with Windows 8 performance on the platform. There's also the fact that Windows 8 isn't expected out until the end of next year, at which point we'll likely see an upgraded successor to Bulldozer.

Posted Image

 

I know I'm quoting Anand Lal Shimpi here. But he's the only one I've come across that has the image from AMD. Funny... he didn't run power consumption tests.

 

You can see there, that Win7 is scheduling threads on a module basis.. not core basis. So Full modules aren't being completely gated and shut down. This has negatively impacted power consumption.. and in a little way, performance.

Share this post


Link to post
Share on other sites

Gotta say I am disappointed. I knew single threaded performance wasn't going to be great but to be less than their current offering caught me off guard. Even if it were 10% up in single threaded over Phenom II's I could accept it but this is a hard pill to swallow.

 

It's quite random how well it performs in various tests too, some it can best the 2600k many the 2500k then others it only just beats the Phenom II's or worse not even that.

 

Power consumption is disappointing too. But, with the young 32mn GloFo process that may improve more yet in future releases. The path AMD have taken with the many cores approach over SMT will imo always lead to higher power consumption in the end so that's something else to keep in mind.

 

 

Soras point about it supposed to go up against Nehalem is valid. In that respect it does admirably. Unfortunately they had to spin off GloFo in that time frame which I suspect hindered the release schedule. In any case with Sandy Bridge to go up against today it's quite lack lustre in some departments.

 

Like hell. Architecture innovations go hand in hand with silicon process. Both need codeveloping simultaneously before making a sample. Half of Intels release strategy is the silicon process alone. Can't believe you of all people would overlook that.

 

Besides which AMD has planned upgraded modules for three more releases... And fd-soi research promises guaranteed architecture shrink to the 22nm node

 

How does that sound to you?

This here:

Posted Image

is exactly what I want to start seeing. And executed in a timely manner. With Intel on a 2-3 year "tock" cadence, a 1 year cadence would be nice indeed and shows the predatorial aggressiveness their top man was talking about.

 

Of course the silicon provides half the solution but I was, as mentioned, talking about the architecture specifically. Llano is the perfect example of an architecture that has been taken to process nodes beyond what it was really intended for.

The silicon will go at whatever pace GloFo and it's allies can produce and AMD can't alter that, but they can focus more on the architecture now that the foundry is spun off.

 

The picture above is exactly what I want to see happen down the track for them to remain competitive.

 

If the performance is there, having slightly higher power consumption with an older/immature process won't worry end users too much.

 

 

I've just got to decide whether or not to go ahead with a board and RAM upgrade, and get Bulldozer later down the track (unlikely at the moment), or go with the norm that is a 2500k system (also unlikely due to funding)

 

Going from a Pentium 4 to an Athlon 64 X2 to a Phenom II were big steps, and I was hopeful that would happen also with Bulldozer

 

EDIT: Might start a thread on that later, as I don't have a clue what to do

What about the FX-4170 if you're just after single threaded performance? 4.2GHz base clock.

Or, seeing as those are based on the same 4 module chip (with disabled modules), wait until a pure 2-3 module chip is released? That'll surely bring better power and higher again clock speeds. The manufacturing process will be more mature then too.

 

 

OMG. Retarded "reviewers" thinking that superpi/hyperpi is a CPU performance metric, when all it is is x87 instruction optimising performance. Something Intel made superfluous by introducing SSE instructions.

 

People who care about superpi results... Don't know whats going on.

Yeah but its a number man and smaller or bigger numbers mean better performance daaaaaaaar. my puter gets 8 seconds yours gets 11 seconds your puter is bad im pro.

 

Oh no. Not 80 bit precision instruction. I guess I'll just have to be content wick pwning all Intel CPUs in 256bit AVX, FMA4, and, XOP optimized benches and apps. :(

 

Damn your proness sir. Damn it to hell.

 

lol, don't start this up again.

 

It's a part of the CPU that still gets used (physx). Yes it's old and surpassed in many ways but it's there. Just so happens it's benchable, so people are gonna bench it. It's what enthusiasts do, it's another metric to measure it by.

 

 

EDIT: Regarding the Win7/8 scheduler, the image above that in the Anand article sceptre I found interesting:

Posted Image

Edited by mark84

Share this post


Link to post
Share on other sites

Hopefully no one tries doing Bulldozer benchmarks using XP given even Win7 isn't doing it right ... LOL

Share this post


Link to post
Share on other sites

Gotta say I am disappointed. I knew single threaded performance wasn't going to be great but to be less than their current offering caught me off guard. Even if it were 10% up in single threaded over Phenom II's I could accept it but this is a hard pill to swallow.

 

It's quite random how well it performs in various tests too, some it can best the 2600k many the 2500k then others it only just beats the Phenom II's or worse not even that.

 

Power consumption is disappointing too. But, with the young 32mn GloFo process that may improve more yet in future releases. The path AMD have taken with the many cores approach over SMT will imo always lead to higher power consumption in the end so that's something else to keep in mind.

Yeah. All those different reviewers and all those different outcomes. The Anandtech review was a joke though, no matter how touching his final words were.

 

I thought the guru3d review was fairly good, covered much more... and went into more depth, as they usually do, coz they're proper nerds. In their tests the FX helped up quite well, and does kind of justify itself. The single threaded performance was I grant you... not great. But it doesn't suck either. It'd be more than adequate for anyone's asking.

 

Like hell. Architecture innovations go hand in hand with silicon process. Both need codeveloping simultaneously before making a sample. Half of Intels release strategy is the silicon process alone. Can't believe you of all people would overlook that.

 

Besides which AMD has planned upgraded modules for three more releases... And fd-soi research promises guaranteed architecture shrink to the 22nm node

 

How does that sound to you?

This here:

Posted Image

is exactly what I want to start seeing. And executed in a timely manner. With Intel on a 2-3 year "tock" cadence, a 1 year cadence would be nice indeed and shows the predatorial aggressiveness their top man was talking about.

 

Of course the silicon provides half the solution but I was, as mentioned, talking about the architecture specifically. Llano is the perfect example of an architecture that has been taken to process nodes beyond what it was really intended for.

The silicon will go at whatever pace GloFo and it's allies can produce and AMD can't alter that, but they can focus more on the architecture now that the foundry is spun off.

 

The picture above is exactly what I want to see happen down the track for them to remain competitive.

 

If the performance is there, having slightly higher power consumption with an older/immature process won't worry end users too much.

Yes... but what I'm getting at, is that AMD still have to work with the silicon process to get their design working properly. Released products are called respins, where they've massaged the architecture and the silicon to work better. And no doubt GF employees help the engineers with this process... in between them testing yields and transistor specs on little chips they fab and play around with.

 

EDIT: Regarding the Win7/8 scheduler, the image above that in the Anand article sceptre I found interesting:

Posted Image

Not much of an improvement. But they are still just games. Doesn't really give us a good point of view to work with. If he wasn't prepared to do a variety of some of the other tests, he shouldn't have bothered.

 

In the end. There is hope for this line of processors. Respins. Process maturation. Windows 8 scheduler enhancements. And finally anything AMD can do in a driver to iron out these teething troubles.

Share this post


Link to post
Share on other sites

Where's the like button? That was a very informative post. Thanks :-)

couldn't help but post this.

Posted Image

Share this post


Link to post
Share on other sites

First Australian retailer to advertise the FX processors (according to staticICE)

 

http://www.itsdirect.com.au/products/CPU-A...b_category/181/

 

For an outrageous $360 And no stock till November. "Tell him he's dreamin'". So if the process matures in that time... (and here's hoping we don't get already fabbed parts shipped then) Then we may have better proc's than have been tested. All I can do is be extremely optimistic at this point. But if I'm not overclocking. I can live with the power consumption.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×