AMD 45nm Deneb Pictures, CPU-Z and Super Pi Results

[Yokozuna]

http://en.hardspell.com/doc/showcont.asp?news_id=3721

Looks like K10.5 is about 12% faster than (my) K10 in Super Pi, with a 1m time of 25.109s at 2.8GHz compared to my 9850BE's time of 28.110s at 2.8GHz, with the same HT speed (1.8GHz), same memory speed (DDR2 800) and similar timings as the Deneb ES, though my NB speed is 2.0GHz, compared to the Deneb's 1.8GHz.

[gillll]

nice to see 3ghz 45nm cores from amd

[ghost101]

Are these still power hungry though? The 65nm phenoms didnt compare very well against their g0 counterparts. I wonder how these compare against yorkfields.

[breakfromyou]

1) site doesn't work. seems to be down?
2) if cpu-z is reporting the core voltage accurately, lol.
3) L3 cache was bumped from 2mb to 6?! maybe that explains the performance increase. hmm...

[Hornet331]

nice to see Deneb it gets faster, but will it be fast enough against nehalem?

also i seconed ghosts comment, powerdraw would be nice to know.

[Calmatory]

1) site doesn't work. seems to be down?
2) if cpu-z is reporting the core voltage accurately, lol.
3) L3 cache was bumped from 2mb to 6?! maybe that explains the performance increase. hmm...

I doubt it gives any performance in SuperPI.

Really, wish they ditched the K8 arch and created something new. Now they need to do it anyway, as it seems that K8 is dead and buried, and has been neraly 2 years or so.

Even though they had Superior Arch compared to C2D, they would still be behind due to the manufacturing process. Lower yeilds and higher costs, more power hunggry chips etc.

They can't compete in performance. They can't compete in manufacturing costs. They can't compete in TDP. What can they really do, other than wish Nehalem to fail and delay big time and Bulldozer to be a huge success?

[Macadamia]

L3 cache size should NOT improve performance on super pi.

They probably improved the L2/L3 latency, the core could have some faster workarounds, etc.

[BulldogPO]

Nice to see AMD's 45nm CPU, even tough it is too late to really compete with Intel.

[awdrifter]

This is some good news for AMD. But it's way to late, Phenom should've released like this at the beginning of this year.

[Seraphiel]

This is a pleasant surprise. Didn't think we'll see Deneb ES so soon. I hope that the TDP is low, prices likewise, and that there will be a 3 GHz SKU.

[mAJORD]

Almost new AMD 1M WR

[nfm]

I wonder how many pipelines it has. If there were able to shorten them and keep the same or higher clockage then

[Tony]

I doubt it gives any performance in SuperPI.

Really, wish they ditched the K8 arch and created something new. Now they need to do it anyway, as it seems that K8 is dead and buried, and has been neraly 2 years or so.

Even though they had Superior Arch compared to C2D, they would still be behind due to the manufacturing process. Lower yeilds and higher costs, more power hunggry chips etc.

They can't compete in performance. They can't compete in manufacturing costs. They can't compete in TDP. What can they really do, other than wish Nehalem to fail and delay big time and Bulldozer to be a huge success?

it is new, K10 is nothing like K8 really and this respin is new again.

Intel core 2, based on P3, everyone loves that though Nahelem yes its totally new but thats just really a move towards what AMD are doing but much faster.

[KeZzZu]

This is just awesome ! i hope they really lowered latencys on l2/ l3

[MomijiTMO]

Nice chip there. Now we have to wonder if retail chips are this fast/good.

[Cooper]

Although SPi ain't the benchmark to show performance on AMD processors it's still nice to see 3.4GHz run.

[Carfax]

If Deneb outpeforms my current Yorkfield processor and overclocks well, I'll pick one up at the end of the year rather than getting a Nehalem based system. I figure I'll wait till Nehalem is on the "tock" phase

[gOJDO]

The link doesn't work for me. It'd be nice if anyone can re-host and post the pictures.
Anyway, it is nice to see ES floating around. That means the release date is not too far away.

The L3 size improves the performance in Super Pi for sure, but I doubt that 12% are delivered from the L3 size alone. Since I can't open the page I wonder what is the NB & L3 frequency. I hope for more benches soon.

I wonder how many pipelines it has.

It has 3 pipelines, just like Agena. As for the pipeline length I'm pretty sure it is same as on Agena, 12/18 ALU/FPU stages.

If there were able to shorten them and keep the same or higher clockage then

Shortening the pipeline requires serious and complex architectural redesign(which is very unlikely to happen on a brand new process). It will make a lot of troubles to the engineers and I doubt it'll make any benefit since K10 relatively has a short pipeline(compared to Core2 and Nehalem). What K10 needs is higher core, L3 and NB clocks, lower cache latencies, wider cache and some architectural improvements.

[v0dka]

Thanks for these results. Was totally not expecting any.

Looking at Phenom development with the SB750, I have my hopes up for Deneb overclocking to the >3.5Ghz range for daily use. Just hoping that power consumption will be low to finally make it an easy, viable alternative to Core.

Then there's only the bad image to fight, I mean who cares about 15% extra performance if the price is right, when GPU or HDD are bottlenecking your system almost exclusively (exclude benchers and crunchers here). If youre lacking 15% performance and at the same time pulling 50w extra it's a whole different scenario though.

Nehalem is a different ballpark. When mainstream Nehalem enters AMD will have to do something good or accept defeat.

[roadie]

This is great news. I look forward to seeing how the chips develop.

[Shintai]

Nice results sofar. A guess would maybe be corespeed L3? The async nature is a major culpit on 65nm K10s.

[Calmatory]

it is new, K10 is nothing like K8 really and this respin is new again.

Intel core 2, based on P3, everyone loves that though Nahelem yes its totally new but thats just really a move towards what AMD are doing but much faster.

Well, isn't P3 based on P2 anyway? And P2 is based on Pentium Pro which is based on original Pentium? Larrabee having multiple smaller cores which are very alike to original Pentium cores?

So whats the problem with AMD, if not the architecture? Too low yields? Too high costs? Too bad fab process? Whatever it is, it's killing AMD. :|

[Macadamia]

Nice results sofar. A guess would maybe be corespeed L3? The async nature is a major culpit on 65nm K10s.

The FIFO is there to stay, w/o it you'd need major reworking and BIOS flashes all the way.

First post- Deneb 1.8Ghz Agena 2Ghz

They should've improved the logic of L2 and L3- my best guess, the 65nm versions are subpar compared to C2. Now they might actually have something in the vicinity, hampered by their execution units and prefetch.

[Shintai]

The FIFO is there to stay, w/o it you'd need major reworking and BIOS flashes all the way.

You will be required to flash your BIOS anyway. Else it wont even start.

[Shintai]

it is new, K10 is nothing like K8 really and this respin is new again.

Intel core 2, based on P3, everyone loves that though Nahelem yes its totally new but thats just really a move towards what AMD are doing but much faster.

With that view on it, everything is based of the previous then. In short, they are all just improved 4004.