Not going to amortize those years of R&D selling them at $99.
SNB-E is big pile of meh so AMD will have some breathing room.
SNB-E is big pile of meh so AMD will have some breathing room.
Poll: Who do you think will be faster? You do not have permission to vote in this poll. |
|||
Sandybridge | 32 | 82.05% | |
Bulldozer | 5 | 12.82% | |
Same | 2 | 5.13% | |
Total | 39 vote(s) | 100% |
* You voted for this item. | [Show Results] |
Bulldozer vs. Sandybridge
|
09-13-2011, 01:12 PM
Not going to amortize those years of R&D selling them at $99.
SNB-E is big pile of meh so AMD will have some breathing room. 09-14-2011, 09:26 AM
"Normally if given a choice between doing something and nothing, I’d choose to do nothing. But I would do something if it helps someone else do nothing. I’d work all night if it meant nothing got done."
-Ron Swanson "I shall be a good politician, even if it kills me. Or if it kills anyone else for that matter. " -Mark Antony
Couldn't be bothered to even run SuperPi?
Another leaked bench loses to Ph2x6 in Cinebench.
new IVY Bridge Reports are saying upto 20% increased Performance with the same Clock to sandybridge...
due to 22nm and tri-gates lower temps-lower voltages... = higher clocks... sry amd fanboys... low budget is still low budget... and i give a shit to the world record of amd...
EDIT by neobrain: that pic was kinda annoying..
EDIT by dannzen: don't fuck with my sig EDIT by neobrain: yet, I will keep doing it EDIT by ???? : A WILD DACO APPEARS EDIT by [SS]: Hey guys, what's going on here? EDIT by dannzen: Gotta Catch 'em All! EDIT by ???? : WILD DACO BROKE FREE FROM MASTER BALL (09-15-2011, 06:45 PM)lamedude Wrote: Couldn't be bothered to even run SuperPi? The chip is still under heavy nda, take those dodgy "leak" with a grain of salt. As for ivy, well, AMD has working Piledriver cores already, and it's supposedly have higher IPC than BD, so it's not like all hope is lost. http://www.anandtech.com/show/4813/look-...deo-at-idf btw, I love AMD naming scheme 09-16-2011, 07:08 AM
(This post was last modified: 09-16-2011, 07:13 AM by NaturalViolence.)
Well that benchmark (assuming it is real) has the bulldozer chip running at 3.1 GHz. Which is really low for bulldozer (8170 FX can turbo up to 4.5GHz). And at that clock rate the sandy bridge cpu running at 3.3GHz is beating it by about 9% in performance, which means performance per clock is only about 6% higher. If indeed performance per clock of sandy bridge is only 6% higher than bulldozer then bulldozer will win in general against non-HT sandy bridge chips since bulldozer chips are usually clocked about 15% higher than sandy bridge chips on average (if you look at their fall lineup). Those are not bad results at all for AMD. Normally Intel desktop chips utterly destroy amd chips in superpi (that particular benchmark always tends to favor intel chips significantly for some reason, probably has something to do with SSE).
I don't believe AMD would be confident enough to price most of their high end bulldozer chips in the $225-300 range in between the sandy bridge 2500k and 2600k if they didn't perform in between the two. AMD has always been famous for its competitive pricing schemes.
"Normally if given a choice between doing something and nothing, I’d choose to do nothing. But I would do something if it helps someone else do nothing. I’d work all night if it meant nothing got done."
-Ron Swanson "I shall be a good politician, even if it kills me. Or if it kills anyone else for that matter. " -Mark Antony Quote:Well that benchmark (assuming it is real) has the bulldozer chip running at 3.1 GHz. Which is really low for bulldozer (8170 FX can turbo up to 4.5GHz). And at that clock rate the sandy bridge cpu running at 3.3GHz is beating it by about 9% in performance, which means performance per clock is only about 6% higher.assuming it's real, you do realise that you're comparing 4C/4T with 8C/8T. that mean BD's per core performance only about half of intel's at the same clock (see comparison of i5 2400 running at 3.1, again assuming this was real). with that extra core BD surely can compete with intel, but only at multithread apps (rendering, encode using x264 multithread, etc) Quote:Normally Intel desktop chips utterly destroy amd chips in superpi (that particular benchmark always tends to favor intel chips significantly for some reason, probably has something to do with SSE).superPi is singlethreaded. it won't take advantage from multicore cpu. that's why intel have the upper hand, they have better performance-per-clock-per-core than AMD
09-16-2011, 08:34 AM
(09-16-2011, 07:08 AM)NaturalViolence Wrote: Well that benchmark (assuming it is real) has the bulldozer chip running at 3.1 GHz. Which is really low for bulldozer (8170 FX can turbo up to 4.5GHz). And at that clock rate the sandy bridge cpu running at 3.3GHz is beating it by about 9% in performance, which means performance per clock is only about 6% higher. If indeed performance per clock of sandy bridge is only 6% higher than bulldozer then bulldozer will win in general against non-HT sandy bridge chips since bulldozer chips are usually clocked about 15% higher than sandy bridge chips on average (if you look at their fall lineup). Those are not bad results at all for AMD. Normally Intel desktop chips utterly destroy amd chips in superpi (that particular benchmark always tends to favor intel chips significantly for some reason, probably has something to do with SSE).Cinebench benchmark you mean? Your logic is false here since it is a multithreaded benchmark, and the performance of a Bulldozer 8 core processor (most cores they will have when released) in it is about equal to a similarly clocked four core/four threads(without HT) Sandy Bridge processor. Which means that individual core performance is lower than your estimates . I hope the benchmark is fake though... 09-16-2011, 08:57 AM
(This post was last modified: 09-16-2011, 09:01 AM by NaturalViolence.)
Quote:Cinebench benchmark you mean? Your logic is false here since it is a multithreaded benchmark, and the performance of a Bulldozer 8 core processor (most cores they will have when released) in it is about equal to a similarly clocked four core/four threads(without HT) Sandy Bridge processor. Which means that individual core performance is lower than your estimates Wink. I hope the benchmark is fake though... That depends if it uses integer or floating point calculations, and if it's a synthetic benchmark then it is probably mostly FP calculations. Each bulldozer module has 2 ALUs and a shared FPU. A 4 module bulldozer chip has 8 ALUs and 4 FPUs, not 8 ALUs and 8 FPUs which would be considered a true 8 core cpu. As I mentioned earlier in the thread (I think) AMD is using the term "core" very loosly here. I would not consider their definiton of core correct, what they are considering a "module" is what I consider a core. And by that different counting method is would be considered a quad core cpu. Normally each "core" in a cpu has one ALU and one FPU (snady bridge for example has 4 ALUs and 4 FPUs). It's just a different way of counting/looking at it. You could consider it 4 cores with twice the ALUs per core, or you could consider it 8 cores with a shared FPU for each pair of cores. AMD uses the second since it allows them to put "8 cores" on the box and makes them sound better. Just like how nvidia and ati count SPs differently on their chips, 1600 SPs on an ati gpu is not the same as 1600 SPs on an nvidia chip since the two brands have a different idea of what they consider an SP. I would never consider a 4 module bulldozer chip to be a true 8 core cpu. Don't buy into the marketing bullshit AMD is pulling on this one.
"Normally if given a choice between doing something and nothing, I’d choose to do nothing. But I would do something if it helps someone else do nothing. I’d work all night if it meant nothing got done."
-Ron Swanson "I shall be a good politician, even if it kills me. Or if it kills anyone else for that matter. " -Mark Antony 09-16-2011, 04:32 PM
Well as long as it perform as good or better for its price, I don't care how they want to market it.
I'm especially interested in x264 and emulator performance. Have seen some positive figure on x264 benchmark of almost official (still a bit dodgy though). |
« Next Oldest | Next Newest »
|