Pre-release Tests Conducted on AMD Denreb

[btarunr]

AMD Deneb is the code-name for the 45nm quad-core CPU which AMD plans to release soon. Chinese website ITOCP got their hands on two engineering samples. They used these samples at various clock-speeds set by altering the FSB multiplier and Vcore voltage. These chips were then subjected to rounds of Super Pi 1M benchmark. The results look rather luke-warm compared to what we saw of the Intel Bloomfield chips recently. The Deneb CPUs were supported by an AMD RD790 motherboard and 2 GB of DDR2 800 MHz unganged memory, running at timings of 5-5-5-18. The Phenom X4 Deneb 45nm will feature 6 MB of L3 cache apart from the usual 512 KB L2 caches dedicated to the cores.



Source: ITOCP

[kenkickr]

Alex, I'll take the 3.4Ghz for $235!! JK. If I read that right 3.4Ghz @ almost 1.6v!! Wonder how they are cooling it. Anyways can't wait to get my grubby hands on one.

[tigger]

That looks a bit voltage hungry,my E6750 does 3.6ghz at 1.41v/3.2ghz at 1.35v.

[[I.R.A]_FBi]

my e6400 does 3.2 @ 1.225V

[Judas]

Originally Posted by tigger69

That looks a bit voltage hungry,my E6750 does 3.6ghz at 1.41v/3.2ghz at 1.35v.

Lol your's is a dual core this is a quad, mind you it does seem quite power hungry

[DOM]

Originally Posted by Judas

Lol your's is a dual core this is a quad, mind you it does seem quite power hungry

look at specs that is alot of volts

[Judas]

Originally Posted by DOM

look at specs that is alot of volts

lol very funny

[Wayward]

Pre-release Tests Conducted on AMD Denreb

Denreb?

[newtekie1]

Those super-Pi times are pretty bad, I would be embaressed.

[Basard]

seems like they could have used a better system setup.... unganged memory, that is not dual channel, right? and crappy timings...

im not sure if that has much impact on superpi, but it does on everything else...

[newtekie1]

Originally Posted by Basard

seems like they could have used a better system setup.... unganged memory, that is not dual channel, right? and crappy timings...

im not sure if that has much impact on superpi, but it does on everything else...

That is Dual Channel. And 5-5-5-18 isn't terrible timing for DDR2-800, in fact they are pretty standard timings.

[kenkickr]

unganged is best used if you are running alot of multi core apps since it will allow the Phenom to use both memory controllers. Unganged is 64-bit dual channel, ganged is 128-bit dual channel.

[postumus]

these are really early "results", so let's just wait and see. also it's pretty pointless to compare a dual core intel vs a quad core+IMC amd in terms of power consumption.

[PVTCaboose1337]

I got quite a jolt from looking at all those volts that the CPU needs to run. Looks like I'll be needing a new PSU.

[jydie]

Originally Posted by newtekie1

Those super-Pi times are pretty bad, I would be embaressed.

I thought Super-Pi was not "multi-thread friendly"? Is there a new version that actually benefits from having more the one core in your CPU? If not, then Super-Pi is not really a good benchmark for the current multi-core CPU's.

[phanbuey]

wow... poor amd...

[newtekie1]

Originally Posted by jydie

I thought Super-Pi was not "multi-thread friendly"? Is there a new version that actually benefits from having more the one core in your CPU? If not, then Super-Pi is not really a good benchmark for the current multi-core CPU's.

It is a guage on how a single one of those cores performs, which we can then use to get an idea on how all 4 will perform. If 1 core performs like crap, adding 3 more crappy performing cores just gives you a quad-core processor that performs like crap compared to other quad-core processors.

[jydie]

Originally Posted by newtekie1

It is a guage on how a single one of those cores performs, which we can then use to get an idea on how all 4 will perform. If 1 core performs like crap, adding 3 more crappy performing cores just gives you a quad-core processor that performs like crap compared to other quad-core processors.

OK... thanks! That makes sense. As long as people realize the fact that it is only measuring one of the cores, then I can see it's use for benchmarking. So, then you should be able to run 4 Super-Pi tests at the same time on a quad-core, and they would finish MUCH faster then running 4 tests on a single core CPU... right?

[Darren]

Originally Posted by newtekie1

It is a guage on how a single one of those cores performs, which we can then use to get an idea on how all 4 will perform. If 1 core performs like crap, adding 3 more crappy performing cores just gives you a quad-core processor that performs like crap compared to other quad-core processors.

I don't see why you are concerned about Super PI results. Who cares about synthetic benchmarks as long as the processor performs in real world tests and is cheaper than the competition. The Q6600 would beat the 9850 BE in Super PI yet would loose in majority real world tests.

Originally Posted by btarunr

It's not like synthetic-benchmarks aren't indicative of real-world performance at all.

Whatever. Super PI has no baring on the results one may achieve in a game or in a spreadsheet or any other application one uses on a day to day basis.

Originally Posted by phanbuey

ERM.... I would LOVE to see those tests... Here's some that say the exact opposite:
http://www.legitreviews.com/article/735/1/

9850 can barely touch the Q6600 (at stock clocks), and only matches it in 2-3 tests. BTW the 9850 gets KILLED in the Excel spreadsheet tests.

The Q6600 and 9850 BE are equivalent processors, my point is that the Q6600 would win in Super PI but a huge margin yet but won't win in other benchmarks (mostly real world) by the same margin, and hence why SuperPI isn't the best indication of performance.

Originally Posted by yogurt_21

damn this is crazy amd needs t lose the phenom ide and move onto something better and totally redesigned. while the performance ios definetly higher than current phenoms super pi's it isn't anywhere near what current or future intel quads.

Again who cares about SuperPI as long as the CPU's perform well in the game your playing, or the application you use and the performance decent and is cheaper than the competition (Intel) that's all that matters not some silly score on some silly program.

[Tatty_One]

Originally Posted by tigger69

That looks a bit voltage hungry,my E6750 does 3.6ghz at 1.41v/3.2ghz at 1.35v.

My E8200 does 4gig on 1.3V

[btarunr]

It's not like synthetic-benchmarks aren't indicative of real-world performance at all.

[phanbuey]

Originally Posted by Darren

I don't see why you are concerned about Super PI results. Who cares about synthetic benchmarks as long as the processor performs in real world tests and is cheaper than the competition. The Q6600 would beat the 9850 BE in Super PI yet would loose in majority real world tests.

ERM.... I would LOVE to see those tests... Here's some that say the exact opposite:
http://www.legitreviews.com/article/735/1/

9850 can barely touch the Q6600 (at stock clocks), and only matches it in 2-3 tests. BTW the 9850 gets KILLED in the Excel spreadsheet tests by intel dual cores.

[yogurt_21]

damn this is crazy amd needs t lose the phenom ide and move onto something better and totally redesigned. while the performance ios definetly higher than current phenoms super pi's it isn't anywhere near what current or future intel quads.

[johnnyfiive]

Awesome...45nm ftw.

[newtekie1]

Originally Posted by jydie

OK... thanks! That makes sense. As long as people realize the fact that it is only measuring one of the cores, then I can see it's use for benchmarking. So, then you should be able to run 4 Super-Pi tests at the same time on a quad-core, and they would finish MUCH faster then running 4 tests on a single core CPU... right?

Yes, running 4 SuperPi tests at the same time would finish much quicker on a quad than on a single core. Also, because of the shared cache, if you ran 4 SuperPi tests at the same time, the times would be slower than if you just run 1. When you run 1, that single test gets the entire cache on the processor for itself. When you run 4, they have to share the L3 cache.

Originally Posted by Darren

I don't see why you are concerned about Super PI results. Who cares about synthetic benchmarks as long as the processor performs in real world tests and is cheaper than the competition. The Q6600 would beat the 9850 BE in Super PI yet would loose in majority real world tests.

I also would like to see the tests proving the 9850 BE beating the Q6600 in the majority of realworld tests. Everything I have seen show otherwise.

And superpi is important because it does give an idea of processor performance. It doesn't give a well-rouned idea, it only shows a single aspect of the processors performance, but it is still important. It shows how good the processor is at pure number crunching.

Originally Posted by Darren

Whatever. Super PI has no baring on the results one may achieve in a game or in a spreadsheet or any other application one uses on a day to day basis.

In the past, SuperPi results have been very good indicators of real world performance. Can you show me proof otherwise?

Originally Posted by Darren

The Q6600 and 9860 BE are equivalent processors, my point is that the Q6600 would win in Super PI but a huge margin yet but won't win in other benchmarks (mostly real world) by the same margin, and hence why SuperPI isn't the best indication of performance.

We are all waiting on these benchmarks to prove this.

Originally Posted by Darren

Again who cares about SuperPI as long as the CPU's perform well in the game your playing, or the application you use and the performance decent and is cheaper than the competition (Intel) that's all that matters not some silly score on some silly program.

We care because it gives us an idea about how the processor performs.