Wolfdale's 5% performance increase over Conroe. Penryn Benched.

[gpippas]

Anand benches Wolfdale against Conroe.

http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3069&p=1

With all the hype I'm surprised by the results of only a 5% increase in performance. Overclocking looks promising though. They say its also not the final stepping. Wasn't too impressed with the power consumption figures. They are good but not great.

I suppose at the end of the day they are good results for what is essentially just a die shrink. Too much hype was the problem.

On another note looks promising for AMD. If Phenom can at least match those figures AMD might stay in buisness after all.

 

[turboflame]

Yea, I remember Intel saying something about a 40% performance increase from the high-k and metal gate materials used in their 45nm processors.

But still, This isn't the final product and Intel could clock these things sky high.

 

[rammedstein]

wow, cool, seems like it could do some extreme oc, cmon, 3.2ghz on 1.168 vcore/

 

[gpippas]

3.2ghz on 1.168V is very good. And thats before the new stepping.

 

[wolverinero79]

The key thing to note here is that the A0 stepping of Wolfdale which was dropped into a non-optimized currently available motherboard had an average of 5% gain over a fully optimized (stepping and mobo wise) Conroe in non-SSE4 benchmarks, as well as a 10% power reduction. Considering that it's essentially the same core architechture, that's not bad. I'm excited to see what (non-stolen/leaked) final silicon will look like. :-D

 

[pausert20]

The key thing to note here is that the A0 stepping of Wolfdale which was dropped into a non-optimized currently available motherboard had an average of 5% gain over a fully optimized (stepping and mobo wise) Conroe in non-SSE4 benchmarks, as well as a 10% power reduction. Considering that it's essentially the same core architechture, that's not bad. I'm excited to see what (non-stolen/leaked) final silicon will look like. :-D

I agree with Wolverinero, this is an A0 Engineering Sample. That is runs this good is amazing. I had heard that the Wolfdales will be launching on a Fab C stepping. Now how many actual revisions between A0 and what is actually release we won't ever know but I bet that we will see improvements in the final released stepping of Wolfdale. It might be only 1 or 2% increase from where we see Wolfdale at now but you know that Intel is putting each stepping and revision of the Wolfdale through their validation labs to come out with a great product.

 

[cb62fcni]

3.2ghz on 1.168V is very good. And thats before the new stepping.

Yea, I think it's pretty clear that this is an evolutionary step, no doubt to allow intel a chance to gain experience with their new high-k/metal gate fabbing prior to designing a completely new micro-arch around it. It'll be interesting to see if Intel markets these around current core speeds, or if they push them much higher than conroe and keep roughly the same TDP. I guess that will depend on whether or not AMD steps up. The fact that there don't seem to be tremendous gains clock-for-clock does give AMD some breathing room, but they need to deliver on it, and fast.

 

[rodney_ws]

I read that review this morning and I came away impressed... the 5% increase is per clock. I think it's safe to say that Wolfdale will scale higher than Conroe... and on top of that, it's using less power.

 

[jaguarskx]

What got my attention a while ago was the 100%+ increase in DivX encoding performance due to SSE4. I wish someone would do a H.264 benchmark.

However it is nice to see that even without SSE4 optimizations, Wolfdale is a healthy 10% faster than Conroe.

 

[the_vorlon]

A few points...

This is a very early stepping... BIOS tweeks, final stepping adjustsments ~~could~~ be worth up to 15% or so. This is not a prediction, you never know, but by way of comparison, the gap between early AM2 athlons and the final steppings were almost 20% clock for clock.

These benches don't test SSE4 - These will offer HUGE speedups when and if software arrives, so this has great potential that may or may not get realized.

Final speeds? - we've seen 3.33 ghz on very (as in year before launch) early steppings, you have to believe that 3.66, maybe 4.0 has got to be in the works, especially with the lower power usage we are seeing.

It's not a quantum leap, but it's better than a kick in the head.

At the end of the day, it's really up to AMD.

If Barcelona rocks, Intel tweeks the process for higher clock speeds at the expense of yields and starts doing heavy cherrypicking of the bins, and we get 4.33 ghz parts as required.

If Barcelona sucks, Intel optimizes for yields at the expense of ultimate clock speeds and we limp along at 3.33 ghz till AMD gives Intel a reason to do otherwise.

 

[shargrath]

mmm...AMD looks to have a half decent chance competing with penryn if they get it out working but henri leaving seems like he may be jumping off a sinking ship. Then again intel could just ramp up the clock speed to a point where AMD cant touch them.

 

[systemlord]

Well it looks like Penryn may not offer gamers that much more performance, a least thats what the early review shows. Running cooler & with less power comsumption thats always a plus. They should OC pretty good though, I might just wait for Penryn to come out before buying a quad core since I'm in no hurry. But time will tell.

 

[wolverinero79]

Excellent points, Vorlon. Intel definitely seems to be in the position of strength currently and for the future. Must be nice to hold all the cards.

Systemlord - games actually seem kind of split - some seem to get great results while others are poor. I guess it depends on what you spend your time playing.

 

[weskurtz81]

I thought the graphs on page one were rather funny. Is Intel coming out of the closet or are they planning to use those on AMD?

Anyway, didn't AMD demo a phenom running over or at 3ghz already? If that is the case, and AMD isn't full-o-crap then we could see some healthy competition when k10 comes out.

 

[yomamafor1]

Yea, I remember Intel saying something about a 40% performance increase from the high-k and metal gate materials used in their 45nm processors.

But still, This isn't the final product and Intel could clock these things sky high.

actually, Intel only said substantial performance improvement over Conroe in some application (SSE4 intensive), while other performance gain is minimal.

I think you mixed up AMD's "40% over highest Intel's offering" with Intel's Penryn.

 

[MU_Engineer]

Anand benches Wolfdale against Conroe.

http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=3069&p=1

With all the hype I'm surprised by the results of only a 5% increase in performance. Overclocking looks promising though. They say its also not the final stepping. Wasn't too impressed with the power consumption figures. They are good but not great.

I suppose at the end of the day they are good results for what is essentially just a die shrink. Too much hype was the problem.

On another note looks promising for AMD. If Phenom can at least match those figures AMD might stay in buisness after all.

I'd be interested in seeing the 3 MB cache parts getting benched to see how much of the gains are due to the 4 MB -> 6 MB cache in the full-cache version. I'd bet it's not all that much, a few percent, but it probably does explain a fair bit of the performance gains over Conroe/Allendale.

The real advantage of these chips is the reduction in die size (less cost for Intel), reduction in power consumption at the same speeds (better to battle AMD with) and a higher possible clock speed without getting to Pentium D EE power levels (gives a chance to sell another $1200 chip to suckers who bought a QX6800, X6800, FX-60, and FX-57.)