AMD Piledriver rumours ... and expert conjecture

Page 27 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
Reynod

Reynod

Administrator
Mar 26, 2007
14,921
16
42,215
We have had several requests for a sticky on AMD's yet to be released Piledriver architecture ... so here it is.

I want to make a few things clear though.

Post a question relevant to the topic, or information about the topic, or it will be deleted.

Post any negative personal comments about another user ... and they will be deleted.

Post flame baiting comments about the blue, red and green team and they will be deleted.

Enjoy ...
 
Sort by date Sort by votes
It's can be done, just need to be willing to volt higher then 1.5, something I'm not willing to do as it can dramatically shorten the life of the CPU.

Also it's easier to OC a two or three core CPU then a four and even harder for a six / eight. The less processor resources you have running the less heat is generated at those resources.

Anyhow, 4.2 is absolutely fine with me. Going to water was about reducing ambient noise from the system rather then overclocking. The two EVGA GTX580 Hydro Coppers I have in the same loop as that 970 are the real heat generators.
Click to expand...

Don't believe in that, you won't see how your CPU goes from 10 years duration to 2 years. You won't even see that time.
 
Upvote 0 Downvote
970BE @ 4.2GHz pushing SLi GTX 580's..
bottleneck much.?

serious question, not being smarty pants..
(well kinda maybe....😀)
Click to expand...


Nope not at all. Haven't run into a single situation where the CPU @100% and the GPU's weren't.

Then again I tend to do lots of 3D gaming, something that doesn't stress the CPU much but doubles the GPU requirements.
 
Upvote 0 Downvote
Don't believe in that, you won't see how your CPU goes from 10 years duration to 2 years. You won't even see that time.
Click to expand...

When we talk about CPU lifetime we're not talking in actual time that it'll be good. CPU's can last way beyond 10yrs, I have an AMD K6-2 450 that I've fitted into an old Dell GXi system that I got working by setting the multiplier to 2.0 (interpreted as 6.0 x 66). IC's can last an incredible amount of time, provided their kept away from damp environments.

We're talking the probability of Electromigration and heat killing your CPU while it's under stress. Running a CPU beyond it's designed voltage limit will increase the rate at which lose particles get moved around, eventually this will cause a transistor's pathway somewhere to lock up and *poof* death of IC. The smaller the pathways the easier it is for them to get blocked.
 
Upvote 0 Downvote
Seems like trinity news kinda died all the sudden. It was all the rage before bulldozer launched and now its like barely anything. The leaked benches are pretty good tho. I think it might take AMD another alliteration of APU to get things 100%. Trinity just seems like its ok but could be better kind of thing from the leaks.
 
Upvote 0 Downvote
I recently bought a HP Pavilion notebook with an AMD A8-3530MX inside it and 6GB of DDR3-1600 memory. Going to see how it handles when it arrives. It's purpose was to be a portable gaming system, no where near as powerful as my desktop but able to at least get by while also being light. And while I love those desktop replacement super gaming laptops, their impossible for me to use on business trips and to lug around. My back will thank me later I know.
 
Upvote 0 Downvote
I recently bought a HP Pavilion notebook with an AMD A8-3530MX inside it and 6GB of DDR3-1600 memory. Going to see how it handles when it arrives. It's purpose was to be a portable gaming system, no where near as powerful as my desktop but able to at least get by while also being light. And while I love those desktop replacement super gaming laptops, their impossible for me to use on business trips and to lug around. My back will thank me later I know.
Click to expand...


I have an HP laptop with A8-3500M processor with Dual GPU, 6 GB of RAM DDR3-1333, a 5400 640 GB HDD,( wish it were a SSD,) and the powerbrick is not too heavy. I can lug this thing around pretty easily! I love it. F@H works great as well,( even with the Dual GPU 😉 .)

send it to me so I can test it out for you..
after I move and not disclose my new location, I'll send it back.. 😗
Click to expand...

:lol:
 
Upvote 0 Downvote
Interesting article to say the least, would have liked it if all 4 "single core" were clocked the same

Conclusion could be interpreted 2 ways, the way they did it, ya, makes AMD look bad.

Alternate conclusion : Its very hard to improve upon an already good architecture (a-64) vs very easy to improve a piece of *** (p-4) after all, it took 3.7 ghz p4 (with immature HT) to roughly match a 2.6 ghz fx-55.
 
Upvote 0 Downvote
Interesting article to say the least, would have liked it if all 4 "single core" were clocked the same

Conclusion could be interpreted 2 ways, the way they did it, ya, makes AMD look bad.

Alternate conclusion : Its very hard to improve upon an already good architecture (a-64) vs very easy to improve a piece of *** (p-4) after all, it took 3.7 ghz p4 (with immature HT) to roughly match a 2.6 ghz fx-55.
Click to expand...


I'll give you that as being a true statement then but as of today can you say role reversal.
It takes a much higher clocked Bulldozer to macth a Sandy Bridge today.
 
Upvote 0 Downvote
the branch prediction works fine and according to them the cache is optimal as is. they can get the timing a cache frequencies up. The CPU would be good once they hit 5 ghz which wouldn't be impossible considering ibm's power5 went up to 6 ghz
 
Upvote 0 Downvote
the branch prediction works fine and according to them the cache is optimal as is. they can get the timing a cache frequencies up. The CPU would be good once they hit 5 ghz which wouldn't be impossible considering ibm's power5 went up to 6 ghz
Click to expand...

Blatantly false (talking about BD).

It's been tested and benched, the Cache latency is absolutely horrible, worse then Phenom II. The hit rate isn't to good either. The CPU is stalling the execution units due to excess flushing.

This is all indicative of a bad branch prediction unit and cache access mechanism. You can keep jacking up the speed, but it's not an efficient method of increasing performance. Fix the branch prediction / cache access first, that will yield immediate performance benefits and increase scaling of that performance.
 
Upvote 0 Downvote
Blatantly false (talking about BD).

It's been tested and benched, the Cache latency is absolutely horrible, worse then Phenom II. The hit rate isn't to good either. The CPU is stalling the execution units due to excess flushing.

This is all indicative of a bad branch prediction unit and cache access mechanism. You can keep jacking up the speed, but it's not an efficient method of increasing performance. Fix the branch prediction / cache access first, that will yield immediate performance benefits and increase scaling of that performance.
Click to expand...
http://www.guru3d.com/article/amd-fx-8150-processor-review/13
the branch prediction is pretty ok, doesn't beat sandy bridge but still good enough. as for the caching, Im just quoting AMD. The probably do need better caches and faster caches but according to AMD the ratio as well as the the speed is optimal.
 
Upvote 0 Downvote
http://www.guru3d.com/article/amd-fx-8150-processor-review/13
the branch prediction is pretty ok, doesn't beat sandy bridge but still good enough. as for the caching, Im just quoting AMD. The probably do need better caches and faster caches but according to AMD the ratio as well as the the speed is optimal.
Click to expand...


So ... according to AMD their CPU is just dandy .... please stop.

Cache Latency has been tested and shown to be absolutely horrible, worse then Phenom II even. This is incredibly bad for performance as any line flush's will take a long time to correct. The only way to make up for that kind of latency is an unobtanium branch prediction unit. If your PBU is awesome, then it can avoid excess line dumps through pre-staging everything.

This does not happen in Bull Dozer. It's apparent through the various benchmarks done across the internet. BD severely lacks performance vs not only the Intel architecture but also the previous Stars architecture. There are only so many ways you can do binary math in transistors. Intel's advantage isn't in it's execution units, as those are pretty much the same one every CPU made in the last two decades. Their advantage is their instruction decoding / branch prediction is amazing, this advantage is further compounded by their cache access mechanisms.
 
Upvote 0 Downvote
We can test it, can't we? Disable all but one core...er, module, and run a benchmark comparison on a single-threaded benchmark (winzip is always a good example).

I'd imagine the FX-6xxx is about equal to intels 9xxx series quads in applications that don't scale beyond 4 cores, based on the fact PII isn't that much better then C2Q to begin with...
 
Upvote 0 Downvote
Well one things for sure, the new piledriver architecture better be a whole new planet better than bulldozer.
When a Pentium G dual core can beat the Quad core FX4100, it doesnt fill me with much hope!
 
Upvote 0 Downvote
Well one things for sure, the new piledriver architecture better be a whole new planet better than bulldozer.
When a Pentium G dual core can beat the Quad core FX4100, it doesnt fill me with much hope!
Click to expand...

There is absolutely no way to confirm that, though I'd be pleased to atleast see some experimental benchmarks, or atleast some proof.

But you can't make a monkey into a lion, thats what I think, the Piledriver won't just jump out and claim to look Ivy in the eye.
 
Upvote 0 Downvote
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom