"Steamroller" could crush "Haswell"?!

[chernobog]

LOL

So it 'may' beat Haswell, in one benchmark, based on your calculations, wow.

http://www.anandtech.com/bench/Product/836?vs=697

Add 15 to 40% to FX 8350 results depending on benchmark, if it can use 1 to 4 cores or even up to 8 cores and threads. Then youl get a rough imagine of FX 8550 vs i7 4770k.

 

[inf64]

So far we have only data that there will be 2M Kaveri. If it ends up being close to performance of FX6xxx series it would be great.

 

[frozentundra123456]

I think someone forgot to check AMDs roadmaps besides making up pixie dust. 2M/4T for desktop only with APUs. And servers will keep using Pilediver.

Yea, is there even an FX 8550 formally announced yet? Seems like we have speculation to the second power. First when, if ever, it will be released, and secondly what the performance will be.

Honestly, I don't know why these outrageous threads promoting amd keep appearing, but I think instead of making one want to use amd they actually promote bag feelings toward amd. I know that is the effect it has on me.

 

[chernobog]

So far we have only data that there will be 2M Kaveri. If it ends up being close to performance of FX6xxx series it would be great.

http://www.anandtech.com/bench/Product/700?vs=699

Kaveri will be faster than FX 6300, that's for sure.

Well I looked at a roadmap and we don't see 6 and 8 core high end server chips so it indicates that there is a gap to be filled, so I would not be on rampage and claiming that there won't be Steamroller FX since is like saying that Skylake from Intel is going to be canceled. That will never ever happen.

@frozentundra123456

They did not announce it, but it does not mean that its canceled or that it does not exist.

Look at server roadmap from AMD, there are no high end 6-8 core server chips so there is a gap. If there was a 6-8 core piledriver chip in the latest roadmap then it would indicate that steamroller was delayed, since there are no 6-8 core server chips I conclude that there will be Steamroller FX.

Keep yelling that there won't be Steamroller FX and when it gets announced and released you will feel like illiterate fool.

 

[boxleitnerb]

"for sure"?
Why do you do this, why don't you wait? A 2M Kaveri would have to be 50% faster per core to beat the a 3M Piledriver. Unlikely imo.

 

[chernobog]

"for sure"?
Why do you do this, why don't you wait? A 2M Kaveri would have to be 50% faster per core to beat the a 3M Piledriver. Unlikely imo.

Steamroller is 15 to 20% faster per core, Steamroller removes the bottleneck/penalty involving multi-threaded performance thus it more performance efficient in multi-threading.

IF Piledriver was to have the same fix as Steamroller involving multi-threading then per core it would be the same but in multi-threaded it would be 20% faster than current Piledriver.

Example:
Piledriver at 5ghz would have same single threaded performance as Steamroller, but in multi-threaded it would be 20% slower than Steamroller because it has the bottleneck/penalty compared to Steamroller that has none.

Bulldozer/Piledriver have a bottleneck, Steamroller has no bottleneck involving multi-threaded performance.

 

[Vesku]

Unless they completely ditched the module concept then there will be some MT penalty even if it's less than 20%.

Unfortunately the gap in the server roadmap suggests that there may be no 4 module FX Steamroller. Also would be odd for there to be desktop Kaveris with more than 2 modules while the server APU is listed as only 2.

I did some guesstimations a little while back on what a 3 module Kaveri might look like performance wise:

http://forums.anandtech.com/showthread.php?t=2321195

With +15% IPC and +10% clockspeed from a FX 6300 (so giving it cache Kaveri probably won't have) it wouldn't quite beat a 3570K in Cinebench 11.5. Although, it would beat it in multi-thread with reduction in the CMT penalty.

 

[crashtech]

It would be great fun to see AMD swipe the performance crown away from Intel again, even briefly. It would be great news for us enthusiasts, since even most Intel fans agree they could use a kick in the arse where enthusiast products are concerned.

I'm planning an upgrade in Q3. Isn't that about the time these new AMD chips will be released? Maybe I'll wait for some real benchmarks of this supposedly fantastic new product.

 

[Arachnotronic]

It would be great fun to see AMD swipe the performance crown away from Intel again, even briefly. It would be great news for us enthusiasts, since even most Intel fans agree they could use a kick in the arse where enthusiast products are concerned.

I'm planning an upgrade in Q3. Isn't that about the time these new AMD chips will be released? Maybe I'll wait for some real benchmarks of this supposedly fantastic new product.

Ivy Bridge-E will make short work of anything AMD could put out in the Enthusiast space.

 

[chernobog]

Unless they completely ditched the module concept then there will be some MT penalty even if it's less than 20%.

Unfortunately the gap in the server roadmap suggests that there may be no 4 module FX Steamroller. Also would be odd for there to be desktop Kaveris with more than 2 modules while the server APU is listed as only 2.

I did some guesstimations a little while back on what a 3 module Kaveri might look like performance wise:

http://forums.anandtech.com/showthread.php?t=2321195&highlight=

There won't be 3 module kaveri, do you people even understand. There is a gap, currently there are no 6-8-10 core high end server chips on AMD's roadmap involving servers and that does not indicate AMD's desktop roadmap.

Just because they currently only listed an quad core APU with Steamroller cores that does not indicate that there won't be Steamroller FX.

This is like the thing with broadwell, everyone expected it to arrive to desktop while it did not. This is the oposite situation.

Everyone claims it won't arrive and then it arrives.

@intel17

"Ivy Bridge-E will make short work of anything AMD could put out in the Enthusiast space."

Overclock FX 8550 to 5Ghz and its close and cheaper

Update:
Ivy Bridge-E is a 6 core, right? So it should score 11.4 to 12.0 points in cinebench 11.5 multi-thread if its clocked at the same clock speed as i7 3770k thus a Steamroller FX 8550 should beat it or be just slightly slower than Ivy Bridge-E in multi-threaded.

 

[Vesku]

Why would they leave a gap in servers between a 4 core APU and 12-16 core Piledriver? Especially if Steamroller has the amazing characteristics chosen for this threads starting Cinebench estimates.

 

[inf64]

That's a good question Vesku.

 

[chernobog]

Why would they leave a gap in servers between a 4 core APU and 12-16 core Piledriver? Especially if Steamroller has the amazing characteristics chosen for this threads starting Cinebench estimates.

I don't know, ask AMD.

Maybe because its too early for them to assume when they could release it, maybe they could release Steamroller FX around the same time as Haswell refresh just to divert attention to them, steal the light from Intel. Strategic move?

@inf64

Nice quote from ShintaiDK in the sign, he probavly feels like a fool right now. wow.

Callouts/personal attacks will not be tolerated
-ViRGE

 

[Ibra]

Overclock FX 8550 to 5Ghz and its close and cheaper

Yeah, with 330 TDP.

 

[chernobog]

Yeah, with 330 TDP.

In your dreams Ibra, FX 8350 at 5Ghz has 220TDP.

Steamroller FX 8550 at 5Ghz should have 180-200 since its smaller and more power efficient, power consumption difference between Piledriver and Steamroller should be 10 to 20%.

By the way TDP does not mean power consumption, i5 3570k had a TDP of 70-80 but it consumes 160-180 watts.

FX 8350 consumes 220 watts and its TDP is 125 watts.

 

[PG]

"Could"... "May".....

All pointless time wasting speculation. I am reminded of something Red Dawn used to say over a decade ago on these very forums. He probably still says it wherever he is. Forgive me if I don't get it exactly correct.

"If my aunt had balls she would be my uncle."

 

[chernobog]

"Could"... "May".....

All pointless time wasting speculation. I am reminded of something Red Dawn used to say over a decade ago on these very forums. He probably still says it wherever he is. Forgive me if I don't get it exactly correct.

"If my aunt had balls she would be my uncle."

hahaha

Well everyone speculates, even when we are doing a simple thing eg you want to ask some woman to a date and then you speculate on how you should proceed to her.

Well maybe you aunt is indeed your uncle... rofl joking here

 

[Sheep221]

In your dreams Ibra, FX 8350 at 5Ghz has 220TDP.

Steamroller FX 8550 at 5Ghz should have 180-200 since its smaller and more power efficient, power consumption difference between Piledriver and Steamroller should be 10 to 20%.

By the way TDP does not mean power consumption, i5 3570k had a TDP of 70-80 but it consumes 160-180 watts.

FX 8350 consumes 220 watts and its TDP is 125 watts.

Dude seriously read what you just wrote.
For first if FX 8350 has 220W TDP it means it has 220W TDP! Nothing else, TDP= max power draw on stock settings
The i5-3570k has 74W TDP which also means 74 not 180! Even if you would OC on nitrogen it still would not draw 180W there are not even boards for that.
This is not leasure center class where you can spread these false claims and your numbers are completely unreal.

 

[chernobog]

Dude seriously read what you just wrote.
For first if FX 8350 has 220W TDP it means it has 220W TDP! Nothing else, TDP= max power draw on stock settings
The i5-3570k has 74W TDP which also means 74 not 180! Even if you would OC on nitrogen it still would not draw 180W there are not even boards for that.
This is not leasure center class where you can spread these false claims and your numbers are completely unreal.

TDP does not mean power consumption... Please look at this link:
http://www.bit-tech.net/hardware/2012/11/06/amd-fx-8350-review/7

i5 3570k on mini-atx motherboard consumes 160 watts while on a regular ATX motherboard it would consume 180 watts thus you fail my friend.

FX 8350 has a TDP of 125 watts not 220 as you want to believe, fanboy.

 

[littlebluesmurf]

All I can safely expect from Steamroller is the 8 core chip being faster in well multithreaded scenarios than the i7. When it comes to singlethread performance steamroller has no hope of catching up with Haswell other than closing the gap a bit.

 

[Mallibu]

@intel17


Overclock FX 8550 to 5Ghz and its close and cheaper

I like how you just compared the imaginary price & imaginary performance of an imaginary CPU that ain't coming any time soon, with a product that's available. This thread sets a new bar on how to do stupid debates based on your imagination.

 

[Phynaz]

RBM?

 

[Arachnotronic]

All I can safely expect from Steamroller is the 8 core chip being faster in well multithreaded scenarios than the i7. When it comes to singlethread performance steamroller has no hope of catching up with Haswell other than closing the gap a bit.

What makes you think that Steamroller will end up being faster than the i7?

 

[chernobog]

I like how you just compared the imaginary price & imaginary performance of an imaginary CPU that ain't coming any time soon, with a product that's available. This thread sets a new bar on how to do stupid debates based on your imagination.

Imagionation? Nope, I just took various available informations and then put them together.

 

[OatisCampbell]

It would be great fun to see AMD swipe the performance crown away from Intel again, even briefly. It would be great news for us enthusiasts, since even most Intel fans agree they could use a kick in the arse where enthusiast products are concerned.

I'm planning an upgrade in Q3. Isn't that about the time these new AMD chips will be released? Maybe I'll wait for some real benchmarks of this supposedly fantastic new product.

I'd love to see AMD take the performance crown with a reasonable TDP as well, would only benefit the market.

Don't think we'll see it beyond highly threaded office type apps, but it would be nice.