AMD Says Bulldozer Is 50% Faster than Core i7

Cookeh · 13 Jan 2011 at 13:09

CAT-THE-FIFTH said:
Exactly.

+1

Personally quite looking forward to this.

opethdisciple · 13 Jan 2011 at 13:19

CAT-THE-FIFTH said:
4 cores with HT.

Again from the article:

"The document cited compared an 8-core processor based on the "Bulldozer" high-performance CPU architecture with a 4-core, 8-thread, Intel Core i7 950 and with a six-core Phenom II X6 1100T CPU, in three different usage scenarios (media, rendering and games)."

Most games only use upto 4 cores AFAIK.

I recently came accross an article that showed 17 games that already utilise 6 cores...... Il try to find it again!

Edit*

Found it:

http://www.pcgameshardware.com/aid,...already-benefit-from-six-cores-CPUs/Practice/

However from a purely gaming perspective SB and Bulldozer are ahead of there times.... the games will bottleneck the cpu rather than the other way round.

djkav · 13 Jan 2011 at 13:29

I think you are forgetting that Bulldozer cores are not like intels cores. You simply cant compare them core for core, it wont work. Who knows, maybe Bulldozer can work a thread on multiple cores.

We wont know how bulldozer works until we get peoples experience using them.

CAT-THE-FIFTH · 13 Jan 2011 at 13:31

djkav said:
I think you are forgetting that Bulldozer cores are not like intels cores. You simply cant compare them core for core, it wont work. Who knows, maybe Bulldozer can work a thread on multiple cores.

We wont know how bulldozer works until we get peoples experience using them.

Yep!

Emlyn_Dewar · 13 Jan 2011 at 13:35

It maybe was near 50% faster on one of the cherry picked benchmarks they performed.
On average, I imagine the CPUs to perform pretty similar. Swings and roundabouts for benchmark wins...

DragonQ · 13 Jan 2011 at 13:36

If their 8-core CPU is 50% faster than an i7 950 but at a similar price point then woo. It doesn't matter if they're "real cores" or modules or whatever.

Again though, overclockability is important. If they can only go to 3.4 GHz that's a problem. They seem fairly power-hungry if a Google translation is to be believed ("125 watts running at 3GHz + with 8-core" from here).

Troezar · 13 Jan 2011 at 13:38

I'm not convinced on the more cores approach for most people and the apps they currently use. I decided to jump on the fast quad bandwagon as by the time it is struggling there might be a real need for more than four cores. It all seems a little similar to early 64bit, it has taken a long time for widespread use and even longer for it to be needed.

I have no concerns buying a 2500k at this point in time

MumboJumbo · 13 Jan 2011 at 13:43

CAT-THE-FIFTH said:
They also tested the Phenom II X6 1100T too. This would mean that it is competing with Sandy Bridge in situations which use less than 4 cores.

Of course the benchmarks have to be confirmed first so I am officially confused at how well or how badly Bulldozer will perform TBH.

I know they did but they arn't making any claims against their own cpu's are they.

Simply put the claims are false in my opinion. Just think about it.

I can imagine a 50% increase in synthetic benchmarks easily for obvious reasons but just think about games. What games do you know that stress out a cpu anywhere near enough to show a 50% increase over what is one of the leading cpu's atm. Sure sandy bridge might be a bit faster but in games its not really noticeable.

If you underclocked the i7 by 50% and ran it up against a sandy bridge you still wouldn't see a 50% overall increase in performance. We don't have games capable of showing such things on cpu's like this.

A sandy bridge can't even show a 50% increase over a q6600 in a game so bulldozer would have to be incredible to show it over a i7.

Thus the truth is its beaten an i7 in a synthetic benchmark and they have added as a side note that they tested it in games too. It's just clever wording, or poorly worded depending on how cynical you are.

CAT-THE-FIFTH · 13 Jan 2011 at 13:46

8igdave said:
I know they did but they arn't making any claims against their own cpu's are they.

Simply put the claims are false in my opinion. Just think about it.

I can imagine a 50% increase in synthetic benchmarks easily for obvious reasons but just think about games. What games do you know that stress out a cpu anywhere near enough to show a 50% increase over what is one of the leading cpu's atm. Sure sandy bridge might be a bit faster but in games its not really noticeable.

If you underclocked the i7 by 50% and ran it up against a sandy bridge you still wouldn't see a 50% overall increase in performance. We don't have games capable of showing such things on cpu's like this.

A sandy bridge can't even show a 50% increase over a q6600 in a game so bulldozer would have to be incredible to show it over a i7.

Thus the truth is its beaten an i7 in a synthetic benchmark and they have added as a side note that they tested it in games too. It's just clever wording, or poorly worded depending on how cynical you are.

The tests were run against BOTH the Phenom II X6 1100T and the Core i7 950.

A 50% increase in gaming performance over a Phenom II X6 1100T would place Bulldozer at around the same performance as a Sandy Bridge Core i5 or Core i7.

MumboJumbo · 13 Jan 2011 at 13:56

CAT-THE-FIFTH said:
The tests were run against BOTH the Phenom II X6 1100T and the Core i7 950.

A 50% increase in gaming performance over a Phenom II X6 1100T would place Bulldozer at around the same performance as a Sandy Bridge Core i5 or Core i7.

You arn't listerning. They say they tested it against a phenom but what else do they say? Nothing. There is no mention of how it compared to a phenom so they can also claim they tested it against a pentium 4 processor. GREAT! So what? They are only making claims against the i7.

You also clearly arn't understanding how these percentages are working. An i7 will max out any game currently quite happily so say its running at 100fps. A 50% increase means that some how another cpu has to be good enough to run it at 150fps. On a game that is already being maxed out by an i7. This just isn't possible. You'd have to jump several generations to be able to get even close to a 50fps improvement.

Also as I keep pointing out it is hidiously clear these claims arn't made on gaming:

"The results varied depending on the tasks run, but, finally, the Bulldozer processor proved itself to be approximately 50 faster than the Core i7 950."

Like I keep saying. It's a claim made on synthetic benchmarks and means nothing anyway even though there is no proof anyway.

You see the word games and make wild claims that its 50% better in games. Truth is these cpu's are still likely falling behind what intel will have bought out around the same time bulldozer comes out.

Emlyn_Dewar · 13 Jan 2011 at 13:58

The problem is that people don't want to read the story, they just want to see the headline, foam at the mouth a little and shout "OMG GAMEZZ!?"

DragonQ · 13 Jan 2011 at 13:59

8igdave said:
I know they did but they arn't making any claims against their own cpu's are they.

Simply put the claims are false in my opinion. Just think about it.

I can imagine a 50% increase in synthetic benchmarks easily for obvious reasons but just think about games. What games do you know that stress out a cpu anywhere near enough to show a 50% increase over what is one of the leading cpu's atm. Sure sandy bridge might be a bit faster but in games its not really noticeable.

If you underclocked the i7 by 50% and ran it up against a sandy bridge you still wouldn't see a 50% overall increase in performance. We don't have games capable of showing such things on cpu's like this.

A sandy bridge can't even show a 50% increase over a q6600 in a game so bulldozer would have to be incredible to show it over a i7.

Thus the truth is its beaten an i7 in a synthetic benchmark and they have added as a side note that they tested it in games too. It's just clever wording, or poorly worded depending on how cynical you are.

It's not all about games you know...

Also, even if it's only 5% faster on average for games but it's the same price as Sandy Bridge (with cheaper motherboards) then it's still a winner, right?

MumboJumbo · 13 Jan 2011 at 14:03

DragonQ said:
It's not all about games you know...

Also, even if it's only 5% faster on average for games but it's the same price as Sandy Bridge (with cheaper motherboards) then it's still a winner, right?

The point is that in a synthetic benchmark you can show some ram to be significantly faster than others and then in the real world you will see maybe 1fps maybe no change at all. To then throw in that they were testing on games also and still claim this same 50% lead clearly shows its not true. Just think about the difference cpu's make in games. It doesn't add up.

I don't doubt these cpu's could be incredible at decoding but thats not what this article is trying to state nor what others in this thread are trying to draw from it.

Further more decoding is hardly the big picture when it comes to a whole range of cpu's. For instance intel brought out the i3/5/7 and 6 core. The question is what is bulldozer bringing? If they are all 8cores then they may not really be of much use to the average user or even enthusiast gamer. Rather limiting them selves so to speak.

But I guess it is possible that they make their cpu's so that it is not software dependent on how many cores are used. But I don't know the details on that.

Nocky24 · 13 Jan 2011 at 14:04

The approximately 50% claim will just be a cherry picked synthetic benchmark, which is most likely multi-core dependant. It's just hype and marketing talk. In gaming you won't see anywhere near 50% gains. AMD only really need to match Intel's offerings (or get close) to compete, so I'm not expecting anything revolutionary over sandy bridge.

MumboJumbo · 13 Jan 2011 at 14:08

I'm expecting it to do what AMD always do. Be slightly under the performance of intels offering, not overclock as well but be cheaper. Thus the enthusiast will likely take intel and the casual person will likely take AMD. Overall AMD will get more sales, particularly in the buisness sector. However in the Pc-uneducated public sector the general public are oftern brainwashed into believing that nvidia and intel are the best. So I guess it probably ends up quite even between the two on that front with those taking amd for their cheapness and those taking intel for their name.

Jamiman · 13 Jan 2011 at 14:12

I hate to point out the obvious, but it says that the AVERAGE improvements of 50% are in three different usage scenarios (media, rendering and games).

Therefore, it might be more than 50% faster in some of these other tasks, and slower in others (such as games that might not be able to deal with that much extra) -.-

Uriel · 13 Jan 2011 at 14:20

Depends which game. There are few, although not many that are still CPU limited on an i7 950.

I'm not expecting much in GPU limted games, where an Athlon II X3 is usually within a few frames of overclocked i7s.

mothermachine · 13 Jan 2011 at 14:23

the reality is that gaming is almost definitely one of the least important things CPU manufacturers are interested in. as has been said, most games can barely make use of 4 let alone 6 or 8 cores, at the moment. and by the time they do there will be 12 and 16 core chips out etc. etc. the real interest for these guys is in server use and digital content creation (DCC) which utilises every core available to the maximum efficiency. and encoding, to an extent as well, perhaps.

makes me lol that people shell out for 6 core CPUs to play WoW and black ops when a dual core CPU for £50 would do just fine for most stuff, still.

i know this is primarily a gaming forum though, so I understand why 95%+ of you guys are concerned mainly with gaming performance. it's no secret that GPU is still king when it comes to game performance. although that gap is closing. intel saw that, which is why we're beginning to see integrated GPUs into chips these days.

AMD have always had the best "value" bang-for-buck but Intel have always held the top spot in terms of raw speed. I hope that can be turned around, because shelling out for top spec Intel machines is a very expensive business. I'd move my whole business over to AMD in a snap if they managed it. (got 8 quad or hex core PCs here, all intel, plus intel laptops).

MumboJumbo · 13 Jan 2011 at 14:32

Jamiman said:
I hate to point out the obvious, but it says that the AVERAGE improvements of 50% are in three different usage scenarios (media, rendering and games).

Therefore, it might be more than 50% faster in some of these other tasks, and slower in others (such as games that might not be able to deal with that much extra) -.-

Considering the game side would likely be in the region of -10% to 10% I think to make it up to a 50% average would be difficult.

Nelly · 13 Jan 2011 at 14:34

50% faster than Lynnfield would make it way better than Sandybridge.

Will take it with a pinch of salt to be honest, I'll buy Bulldozer anyway just for the sake of upgrading and having fun doing it.

The sockets are crap on Intel motherboards - will never buy another Intel rig again till they have a better socket design that doesn't have the risk of breaking the pins and voiding warranty.