AT Benches Nehalem

[Cookie Monster]

Originally posted by: EarthwormJim
I don't think AMD is in as much trouble as people think. Since they can't compete on the desktop front with their current hardware, they're dominating the server market (more profit).

That may continue with the next generation of chips.

Especially when the transition to nehalem from penryn will take longer to filter down to the rest of the desktop market segments. So i expect penryn derivatives to be around for quite some time before we see anything mainstream with nehalem derivatives.

 

[AmberClad]

Originally posted by: Cogman
Do we know yet for sure if intel has really locked the FSB? (IE did they kill overclocking) That would be the saddest thing for me to see.

Haven't we already heard quite some time ago that overclocking will be available on the higher end varieties of Nehalem, but not on the budget variety?

I find it implausible that Intel would completely do away with overclocking, for this reason - they would have some very unhappy Taiwanese motherboard manufacturers on their hands if they were to go that route. I'm specifically referring to those manufacturers that sell high end "enthusiast" overclocking boards that go for >$200-$400.

 

[Cogman]

Originally posted by: AmberClad
Haven't we already heard quite some time ago that overclocking will be available on the higher end varieties of Nehalem, but not on the budget variety?

Last I heard, the only report like this was from fudzilla. But killing off overclocking for the budget variety would essentially kill off overclocking. Very few people actually buy the upper end just to overclock them.

 

[AmberClad]

Well I guess the question is, what's considered budget? The Allendale line? Or everything non-X or non-QX? Without knowing that info, it's too early to draw any conclusions.

 

[Munky]

Originally posted by: JACKDRUID
bad news for AMD...really bad.. losing in both fronts

intel Nehalem beats phenom by 30%
nvidia 9800gtx beats 3870 by 30%...

which mean, in general, if you use an AMD only machine, it will consistantly be 30% slower than any other system.

AMD really need to get their act together...

AMD wont be so badly behind once the 48xx video cards are released. The 4870 is supposedy 30% faster than the 9800gtx, and in the right price range it will have no competition. The gt200 would be priced much higher, while the g92b will not keep up in performance.

 

[semisonic9]

Originally posted by: ricleo2
Damit. I have waited it seems like forever for the new graphic cards to come out. I was all set to build a new gaming machine with a new video card this month. Now I feel like I should wait (again) on this new CPU. Unless this new powerful chip will not effect gaming much. Anybody know or have an opinion?

I'm in the same boat. I intend to buy in the next month or three, once the newer GPUs are out, benchmarked, and reasonably priced. As for GPU/Mobo combos, I'd expect anything you buy now to serve you well for at least a year or more, at which point the Nethalem chips/mobos should widely available and past their early price-gouging days.

~Semi

 

[ricleo2]

Originally posted by: semisonic9

Originally posted by: ricleo2
Damit. I have waited it seems like forever for the new graphic cards to come out. I was all set to build a new gaming machine with a new video card this month. Now I feel like I should wait (again) on this new CPU. Unless this new powerful chip will not effect gaming much. Anybody know or have an opinion?

I'm in the same boat. I intend to buy in the next month or three, once the newer GPUs are out, benchmarked, and reasonably priced. As for GPU/Mobo combos, I'd expect anything you buy now to serve you well for at least a year or more, at which point the Nethalem chips/mobos should widely available and past their early price-gouging days.

~Semi

Thanks for the excellent reply.

 

[kotrtim]

The impressive 40% gain is mainly due to the HT - HyperThreading. QPI benefits servers, not so much for desktop users. If intel would have implemented HT into Core 2 LGA 775, I think a 30-40% is not impossible, but I guess its not gonna happen, HT is exclusively for Nehalem only. Wonder why AMD refuses to implement HyperThreading to its upcoming procs, or is it patented by Intel? Its so difficult to increase clock for clock performance of x86 procs and Intel said its using HT to utilize all the idle transistors, IIRC.

 

[Quiksilver]

I am one happy man right now, I need a computer upgrade soon but I've been putting it off due to this being on the horizon, it's nice to see near a 40% increase clock for clock. I suspecting we may see a 50 to 60% increase clock for clock once all last minute optimizations and chipset bugs are worked out. The only real question that will remain is the ability to overclock them, if you can great, if you can't still great as long as prices are reasonable. End of this year couldn't come soon enough; I recon I'd have about 4/5k to put towards this machine too.

 

[Sylvanas]

Well I know where my next upgrade is headed- and to think performance is expected to be even better with retail motherboards.

 

[GuitarDaddy]

The real question is when will Nehalem become affordable? The early pricing charts (maybe total FUD) have the original release parts ranging from $1300-$2800.

I don't know about you guys, but even a 50-60% performance gain over a $300 chip ain't worth that kind of jack to me. I can't see many enthusiasts (except the X crowd)dropping that kind dough to pimp their gaming/overclocking rig. At these prices the main target would be highend workstations where the production oriented computing power could justify the expense.

 

[LittleNemoNES]

I hope they can delay this so I can feel better about my recent build

 

[Munky]

Originally posted by: kotrtim
The impressive 40% gain is mainly due to the HT - HyperThreading. QPI benefits servers, not so much for desktop users. If intel would have implemented HT into Core 2 LGA 775, I think a 30-40% is not impossible, but I guess its not gonna happen, HT is exclusively for Nehalem only. Wonder why AMD refuses to implement HyperThreading to its upcoming procs, or is it patented by Intel? Its so difficult to increase clock for clock performance of x86 procs and Intel said its using HT to utilize all the idle transistors, IIRC.

Did you see the Divx benchmark?

The DivX test is an important one as it doesn't scale well at all beyond four threads, any performance advantage Nehalem has here is entirely due to microarchitectural improvements and not influenced by its ability to work on twice as many threads at once.
...
Clock for clock, Nehalem is nearly 28% faster than Penryn

I believe the revised caches and the IMC are responsible for most of that improvement, and likewise in other apps, not because of HT. And I don't believe any claims that the FSB memory system in the current quad cores is not a bottleneck in many apps that heavily utilize more than 2 cores, even for desktop users.

 

[biostud]

To me it seems that more and more move to mobile platforms, but the lackluster is of course the 3D performance. With Puma and the external videocard support AMD might have a chance of gaining ground in the mobile gaming crowd. Increasing their sales here while they might loose some ground in the high end desktop market. It will also be interesting to see how the SLI/nehalem feud will end.

 

[bharatwaja]

Hey guys, I am just back after a 3 month trip to Africa and guess what, the first article I read is this one about nehalem.... god it rocks!!!!!

looks like I have truly been in the 3rd world for quire a while now, but am just catching up on happenings....

Originally posted by: Idontcare

Upgrading on the Ticks is probably the better strategy versus upgrading on the Tocks.

Folks who upgraded to Penryn (the tick to Conroe's tock) get better performance clock/clock and lower power consumption than folks who upgraded to the tock (C2D).

Same will be true for Nehalem (tock) versus holding out for the tick (westmere). Since you already have a Q9450 it would seem the better thing to do here is keep with the tick-upgrade cycle and hold-out for Westmere.

That's true... but our upgrades itself usually follow a tick tock routine.... if u set a 3 yr life cycle for a machine (give or take some)... when going intel's way, u naturally have ur own tick tock thing going on.... atleast, thats what happens to me, I got a penryn in the form of qx9650, and my next upgrade would be a tock.... unless of course u wish a 2yr life cycle, but that just doesnt say good economics about you.....

 

[bharatwaja]

man is greedy
We see nehalem, we think of holding out for westmere (or even sandy bridge) to see how that goes....

Will we ever reach a limit?? a limit to what we desire???

naaa! as long as there is some gr8 chip porn here on AT, like this nehalem article, ppl will be happy!

 

[dv8silencer]

"Intel Nehalem To Allow Overclocking, Some Processors Will Even Encourage It"

http://www.tomshardware.com/ne...ntel-Nehalem,5607.html

 

[PokerGuy]

Yowsah.... that's a major improvement, especially considering it's very early and things have not been optimized yet. This is definitely going on my wish list!

 

[Caveman]

So... Is it possible to upgrade to a Penryn MoBo now, and ensure upgradeability to a Nehelem CPU when it's affordable? If so, what MoBo would be recommended? I'm interesting in O/Cing the Penryn and the Nehelem CPU

 

[amenx]

Originally posted by: kotrtim
The impressive 40% gain is mainly due to the HT - HyperThreading. QPI benefits servers, not so much for desktop users.

Yep, everyones excited for what they think Nehalem is going to do for them, but in practical terms I dont think its much. These chips are designed for the server markets, not for gamers or basic single/dual core applications. In multi-threaded applications it may indeed excel, but for the 99% of non-multithreaded stuff and games its no better than a Penryn. Anyone see the single thread bench vs the Penryn?

Frankly I was excited about Nehalem before this AT review. But now less so because now I know it wont be much of a practical improvement over my current CPU for what I currently do and expect to do with it over the next 12-18 months or so. Will wait and see how the desktop chips turn out, but I dont expect much difference.

 

[amenx]

Originally posted by: Caveman
So... Is it possible to upgrade to a Penryn MoBo now, and ensure upgradeability to a Nehelem CPU when it's affordable? If so, what MoBo would be recommended? I'm interesting in O/Cing the Penryn and the Nehelem CPU

Entirely different platforms/sockets, not compatible at all. You'll need a new board when Nehalem comes out.

 

[Hulk]

I will be quite impressed if Nahalem shows a 25% average performance over Penryn. I also bet that further "optimizations" will yield at most 5% improvement. Intel is pretty good at getting things working right from the start so I doubt there is much more left in the chip beyond what currrent testing is showing.

As I said Conroe was such a break through IPC-wise that is was impressive that Intel could actually improve on that from a efficiency standpoint, both in IPC and thermally.

Now they are able to improve Penryn's IPC by perhaps 20% on the low end and 30% or more on some apps? Compared with what the competition is doing I think Intel is doing pretty well.

Hopefully AMD will be able to somehow pull a rabbit of it's hat.

 

[s44]

Originally posted by: dv8silencer
"Intel Nehalem To Allow Overclocking, Some Processors Will Even Encourage It"

http://www.tomshardware.com/ne...ntel-Nehalem,5607.html

Hope it's right, but that article is pretty content-free past the headline.

Originally posted by: Hulk
Hopefully AMD will be able to somehow pull a rabbit of it's hat.

Depends on their 45nm process, I'd think. They were absolutely slaughtered by Intel's process advantage at 65nm.

 

[Nemesis 1]

Originally posted by: Idontcare
Folks this isn't my thread but I'd like to say please keep all your Intel vs AMD arguments at least confined to the context of performance/price.

Will Nehalem slaughter Shanghai? I have no doubt the flagship top-end Nehalem will slaughter the top-end flagship Shangahi...but at what cost to the consumer to acquire that Nehalem flagship versus AMD's?

Only when put into context of price/performance does it make much sense to say things about Intel vs. AMD.

To whatever extent Shanghai/Phenom underperforms Nehalem you can rest assured AMD will price accordingly (maybe not perfectly, maybe not aggressively, but at least accordingly) to account for performance delta's.

Given that we all are not likely to see Nehalem's for the desktop in any form except $1200 extreme edition Bloomfield for another year (or longer if you read any of the speculation websites out there) it really isn't much to get crazy excited about outside of getting excited over exciting tech for exciting tech sake. (I fall into this category, count me +1 excited)

.

The op really doesn't care what you guys post here. Beings how we have moved closer to reallity than speculation. Just lets everyone be good to each other .It doesn't matter if idealisn is differant in fact I encourage it. Just play nice.

Idon'tcare you bring up some interesting points. I would like to add 1 small thing to the price performance . After Nehalem is released .Penryn is still goingto be produced in large numbers. I do believe Penryn will sell ALOT cheaper than now.

After Nehalem for server desktop is released in the 4th qt. It ISN"T competing with pheneom. IT has no competition but Penryn . Penryn will still be competing with 45nm AMD.

This is way differant than the Merom launch. As intels P4 generation was slower than AMDx2. But this time intels last generation is faster than AMDs best. It will be Penryn that puts the hurt to AMD in the consumer market . Nehalem will close the door on AMD in the srver market. Price performance will belong to Penryn after Nehalem and not AMD.

Can you see Penryn qcore selling for under $150 I can.

On the gaming front. I believe the gaming statements are false. NV 280 is going to require more cpu power. Every generation does . It won't belong befor we see GPU reviews and when we see them . Penryn is going to look even stronger than befor . Nehalem may even beable to out perform Penryn in gaming with 20-40% performance increase. 2 weeks we all find out . But nv 280 at 2x the last generation will need more cpu . Same with ATI 4870x2 . These things will need more cpu power.

 

[Idontcare]

Originally posted by: Nemesis 1
On the gaming front. I believe the gaming statements are false. NV 280 is going to require more cpu power. Every generation does . It won't belong befor we see GPU reviews and when we see them . Penryn is going to look even stronger than befor . Nehalem may even beable to out perform Penryn in gaming with 20-40% performance increase. 2 weeks we all find out . But nv 280 at 2x the last generation will need more cpu . Same with ATI 4870x2 . These things will need more cpu power.

Could you imagine the power consumption delta that Intel could score were they to fab performance competitive GPU's on their modern process technology (32nm HK/MG when Larrabee comes out) relative to whatever AMD/Nvidia are getting from TSMC at the time?

Sure Intel has massive performance delta to overcome (as they did with P4 prescott vs K8 X2) but their process technology puts them on a completely different playing field whenever they decide to finally compete in the high-end GPU market.

Those reports on Fudzilla of Larrabee being a 300W monster just doesn't sound right to me, I just don't see Intel's marketing group ever allowing the engineers to create products that can be attacked by the competition for power consumption attributes again.

Originally posted by: s44

Originally posted by: Hulk
Hopefully AMD will be able to somehow pull a rabbit of it's hat.

Depends on their 45nm process, I'd think. They were absolutely slaughtered by Intel's process advantage at 65nm.

From the details AMD has released regarding their initial 45nm implementation (no HK, no MG, no ULK) the gap is only getting larger - both in timeline and in performance. There is no end in sight to rate of increase in the gap, that is what 4X the R&D dollars entitles Intel to have and no amount of innovation can change the fundamentals of this situation.