QFT!
But being an AMD Fanboy is good though:rofl:. Nehalem is evil!
Printable View
It wouldn't matter if it was 20%.
Nehalem improves performance on average over Penryn by 20-30%, and in code where 8+ cores are supported (allowing for SMT to work) the difference can be more like 30-40% or more. Deneb needs to be a solid 10-15% faster than Agena to equal the performance of Penryn.
And no I don't say that with happiness, although I can't wait to get my hands on a Nehalem system. It's not good at all that there is going to be such a wide gap in performance. But there's not much anyone can do about it except hope that AMD is more competitive in the future and with Bulldozer than they are now.
And outside of a very few people, nobody in this thread is an Intel fanboy. Most of the people praising Nehalem right now were laughing at how terrible the Pentium D was and praising the Athlon 64X2 just three years ago.
To all here;
Lets try to keep this friendly ok.
Some people like AMD, some Intel.
Hell I have an old AMD FX51 in the other room that I think is one hell of a machine for a single core but it's time has come and gone...BUT I still keep it because for what it is, it demands respect.
An interesting point for the AMD lovers:
That little single core FX51@2423mhz would outdo my old dual cpu Intel 2400/512/533 machine in DC work... Yes, great respect for that little beast!
BUT today that isn't the case and why I offered LIKMARK to compare his AMD to my Harpertown which I KNOW, not think, is substantially slower clock for clock than the coming Nehalem 8 core Gainstown system.
I generally use Cinebench 10 to show pure cpu power.
A Gainstown at 3000 will out perform my Harpertown at 3574mhz, the max I was able to push the X5482's I had borrowed for 2 weeks on stock volts.
With those X5482's I was able to pull a SP1m time of 13.297s at 3603/12mb/1800( Yes, 450 on FBDimms! ) and thats on a workstation board with FBDimms and you know how badly WS boards do on SuperPi. For comparison the best I was able to get from the Clovers was 15.5s
Now I can tell you that the Harpertown is pretty damn impressive in it's own right. On a daily basis it produces app 40% more than my 8 core clovertown machines and they aren't that bad.
Now taking all this into consideration can anyone wonder why there is so much excitement over the coming Nehalems?
Lets not get lost in another AMD vs Intel battle.
Someone wants AMD, fine by me, buy it. They can use the money coming in but don't tell me they will benchmark with the current Intels let alone what is coming because they just plain won't do it.
That isn't me being a fanboy, thats just plain fact.
You got that right, but you know that's not good enough for some! Nehalem scores are likely to go up as firmware and etc.. is tweaked. We laughed at Hotscott in the Intel part of the forum and no one called it thread crapping!
Nehalem will rock=P Good night guys!
Enough with the QFT for Gods sake.
I would have to disagree. I feel its a very relevant comparison. Especially if you have a thresh-hold of performance, your server needs to meet, as a company that is useful to know. And then it really does come down to price/perf, a 1p intel server vs a 2p amd server with 8 cores theoretically, a comparison between the two is very relevant. Of course google and the like just want max performance, but small business, that have 1 or 2 servers, with specific performance requirements, the dollars and cents of performance vs operating costs on the hair thin margins most web based business operates on, this is a big deal. It will be a real comparison, once nehalem does start shipping, and from what those graphs say... nehalem seems pretty damn efficient, in that regard. So I m not sure how trying to create a realworld comparison as to what will exist in the marketplace is Fanboyism.. I like seeing those numbers to be honest, I know how intel performs against itself :P
5 constructive posts as usual.
Please explain to me the "rules of comparison". I were not aware that I broke your laws of comparing two products. I did not know that if a product were bought on ebay, or was an ES (not released yet) it invalidated the entire comparison. If that rule applies. XS mods need to get their fingers out and moderate all those comparisons we see of new unreleased products. I did not know comparing one cpu setup with a two cpu setup were not allowed either. How would you know the benefits and drawbacks of one cpu system vs a two cpu system if those comparisons were not done?
Please show me where I have downplayed costs. I did not mention costs, this was mentioned by your friend Shintai, and to be honest he downplayed nehalem costs by not mentioning that you'll need a new board and ram.
In every other forum in the internet a comparison of a coming product with products that is out today would be welcomed, even appreciated, but not on XS. Here you get ridiculed by threadcrappers with the mods blessing, wich IMO shouts out where the closet stands. Pardon my bad english, hoping readers understand. That you classify comparing an AMD setup with an Intel setup as threadcrapping is beyond me.
Wonder what the comment would have been like if I compared the nehalem benches with another Intel system like MM is gonna do? Don't think you would have complained as much, probably would have been busy wtf qft bbq'ing all over the place in hope of getting a kind word from one of your heroes.
As for the fanboi remarks, I have my share of Intel setups, so you know where to put those remarks.
On the other hand being marked as a fanboi by you and Shintai almost wants me to bear the mark proudly.
I bought a dual socket system on ebay, and chose AMD because it was cheap, and I find K10 more fun to play with than Core2 and it's FSB. Just as Nehalem will be more fun to play with because of it's advanced arcitechture.
Well thats as much energy I'm gonna use on your remarks. (Much more than what you deserve).
@MM: I'm gonna quote you from deepthought86's thread in the "Forum FAQs and Guidelines" section :
Aww come ON, I don't understand why you mods let them continue degrading the quality of XS.Quote:
I assume you mean Shintai and Donny27?
Two good people. Strong personalities I will grant you
Quote from DeepThought86:
The best suggestion I have heard here on XS, actually it is sig material..Quote:
Just try banning them as an experiment, say for 2 weeks. You can put up a poll to see if people think the News forums are improved or not. But I think you know what the results will be...
@Villa1n: Watch your back and don your flamesuit for not agreeing with the news section mascot.
Where to start..
Yes, you quoted me correctly about Donnie and Shintai.
can I tell you something?
BOTH I actually got to know after asking both to calm down a bit in the threads so I wasn't exactly predisposed to be their "best buddy"
BUT I found both to be pretty sharp guys that knew their stuff.
Shintai's knowledge is at a level that most here couldn't touch but he is still reasonably young( by my perspective at age 56) and the passions of youth do enter his posts.
That doesn't take away from his knowledge, just taint's it with passion.
Now to you and I have no issues with you at all, I think the problem came up from trying to compare your 8 core to the 4 core Bloomfield.
Just not a reasonable comparision at all.
Forget HT, it's not part of the equation in this regard.
4 cores vs 8 just doesn't fly.
Cripes, I remember beating FUGGER's Cinebench score with his Yorkfield at 5500MHz on his cascade with my clovertown on air at 3258..
You can't compare them so this point has to be dropped.
This again, is why I offered to compare my Harper vs your AMD since we don't have a fully functional 8 core Nehalem to use for a comparison.
It would show at the very least how the current AMD stands up against the current Intel core for core and that is a valid comparision.
Now it's 2:50am here, I have a plumber coming at 8:30am so I need some sleep. We can pick this up in 18 hours or so if you wish and I'd prefer it if we all could do this as friends in search of knowledge not people swinging hammers at one another. That benefits no one and just makes my two typing fingers feel like lead at the end of the day.
My best to you all..
Good night..
No one not allowing you to compare with Nehalem, you get the charts from Hornet311 right, all showed up. No one resist Barcelona result you provide ... Until ... (see below) -->
But the problem is you wrongfully compare it with Bloomfield, that's simply different rank with your 2*Barcelona ... That's the first place I provided you the same speed of Gainestown 2.4GHz right ? ... But it has been ignored :yepp: Plus oww, its show 16threads, BUG ?
Still couldn't accept the HT thread thing that showed on OS I think :ROTF: Totally different thing though ...
And, now what ? all become attacking other members personalities ... :up:
Jcornell:
But why is this not an reasonable comparison??! It sounds like a very reasonable comparison to me if you are in a situation where your computing needs could be met by a Quad core Ht enabled system, or if you'll maybe need a dual cpu setup. Wouldn't you want to know wich system suits your needs best?Quote:
But the problem is you wrongfully compare it with Bloomfield, that's simply different rank with your 2*Barcelona ... That's the first place I provided you the same speed of Gainestown 2.4GHz right ? ... But it has been ignored Plus oww, its show 16threads, BUG ?
The 16 thread question was a fair question wich I found out myself when looking closer, np.
Don't know how to talk into you about RANK/HT now ?
Not reasonable ? I think I've chosen WRONGFULLY(COMPLETELY WRONG) to represent this matter.
Okay, okay ... I'll try find some old 2.4GHz Bloomfield chip in the trash and bench it for you, then your Barcelona will shine again, because Bloomfield 2.93GHz 8threads is truely cheating on your Barcelona 2.3GHz*2.
Satisfied? (for you I think) ...
im waiting for those numbers, if you dont mind i'll add them to the charts.
good point, but we dont know how intel gona price gainestown, and it seems that if there is a 2,4ghz version it will be priced according to amd offer, with preium price for the higher clocked versions.
the comparrison is not invalid, but it makes no sense to compare them, cause everyone knows that 8 physical cores are faster then 8 logical cores (4 physical with HT)
what, i dont know what forum you are browsing, but xs still welcomes reviews of upcoming hardware.
maybe that has something to do that mm wouldn't try to compare a single socket system to a dualsocket system, if he would try to compare his harpertown rig to bloomfield, im sure he would earn the same criticism )nothing against you mm ;) )
man everytime i hear this argument i cant help myself, i have to grin... what is so fun on tweaking a system for hours to gain next to nothing?
I have nothing against tweaking, but when it gets excessive and the results are rather weak its not my thing.
I take the "simple and not funny" fsb system every time over the "complex and funny" HT K10 system anyday.
I really hope that ocing doen't get more complicated with nehalem, i dont want to fiedle around for hours to just get some basic overclocking resulst, but if i hunt for max performance i welcome those options.
You're not serious too to compete below system with Phenom right ? :doh:
Bring it on, I'm ready :CTF: ... 4 threads vs 4 threads right ? :clap:
http://i245.photobucket.com/albums/g...neous/Pres.jpg
...
you still have a Pentium D system up and running? wow nice one. :up:
I'm meaning no disrespect, and understand english is not your native language (just like me), but I do not understand what you are trying to say with those two sentences. Could you elaborate?
Ok now I understand.Quote:
Okay, okay ... I'll try find some old 2.4GHz Bloomfield chip in the trash and bench it for you, then your Barcelona will shine again, because Bloomfield 2.93GHz 8threads is truely cheating on your Barcelona 2.3GHz*2.
Satisfied? (for you I think) ...
I'm getting all this sh1t because my dual cpu setup outperformed one single nehalem in one or two benchmarks. My intetions vere not to make nehalem look bad, and I assumed most XS readers understand that there is a difference between 1 cpu -> 2 cpu systems. That's why I did not comment further on it. I am not after making my dual AMD system shine on the expense of a pre launch single nehalem, I am not stupid, I know Nehalem will outperform Barcelona and most possibly Shanghai with a fair margin.
As an enthusiast myself I believed people here would be interested in a comparison between the different systems (as I have stated a lot of times now)
ffs, how many different rigs do you own? hehehe :D
even though i agree with you all that comparing a dualsocket to a singlesocket system isn't that fair - especially if the singlesocket cpu has not the same amount of physical threads. BUT still, even if the comparison isn't fair it's still valid to do such a comparison due to the reasons LIKMARK mentioned: IF a dualsocket barcelona system performs about the same as a singlesocket nehalem system, but only costs a fractional amount i think the comparison is justifiable.
Yes everyone knows that 8 logical threads is faster than 4logical and 4 virtual threads, and I thought it would be nice to see some numbers, not just the general statement that 8 vs 4+4virual is faster. Will anyone tell me why there is no sense of comparing 8 threads/cores with 4+4 threads/cores- ht virtual threads/cores? I find it interesting to compare this, and I believe as said before:that buyers would like to know what meet their computing needs and cost the less.Quote:
But why is this not an reasonable comparison??! It sounds like a very reasonable comparison to me if you are in a situation where your computing needs could be met by a Quad core Ht enabled system, or if you'll maybe need a dual cpu setup. Wouldn't you want to know wich system suits your needs best?
It was not intended as a showdown to crap on nehalem perf, something it obviously appears most readers believe.
It was a response to Donnie27's lame remarks, please read again to get context.Quote:
what, i dont know what forum you are browsing, but xs still welcomes reviews of upcoming hardware.
well I would like to know the difference.Quote:
maybe that has something to do that mm wouldn't try to compare a single socket system to a dualsocket system, if he would try to compare his harpertown rig to bloomfield, im sure he would earn the same criticism )nothing against you mm ;) )
Yes we are all different. As I don't intend to go Xtreme on overclocking (DI/LN2) there is no way for me to for example show my skills in tweaking a intel system as some dude with the right connexions/money craps on your oc with his LN2 system.Quote:
man everytime i hear this argument i cant help myself, i have to grin... what is so fun on tweaking a system for hours to gain next to nothing?
I have nothing against tweaking, but when it gets excessive and the results are rather weak its not my thing.
I take the "simple and not funny" fsb system every time over the "complex and funny" HT K10 system anyday.
Again, people are different :)Quote:
I really hope that ocing doen't get more complicated with nehalem, i dont want to fiedle around for hours to just get some basic overclocking resulst, but if i hunt for max performance i welcome those options.
I would be interested in seeing Netburst/Core2/Nehalem/K8/K10 thread efficiency (or whatever to call it), in a variety of relevant benchmarks. That is because I find such things interesting, not to crap over the guy with the worst performing system, wich obviously is the way to go here in this section.
but thats the thing we dont know, we dont know the pricing of xeon procs, and as i mentioned eralier, if intel releases a 2,4ghz gainestown that slightly undercuts DP opterons in price the whole comparisson between singel socket and dual socket makes even less sense.
Bloomfield is a highend desktop chip and not aimed at servers thats why we have Xeon.
Why should a company buy a bloomfield rig (where the proc costs 1000€ alone) when they can get 2 gainestown procs (lets asume 400€ each) for a lower price which perform a lot better in multithreaded apps (thought they are clocked lower, lats say at 2,4ghz) and simply crush bloomfield?
This I don't get:confused:Quote:
I think I've chosen WRONGFULLY(COMPLETELY WRONG) to represent this matter.
And you can lay off that patronizing act, it does not become you when you display such an obvious lack of english education.
Your teacher probably would have deserved being kidnapped due to failure to learn his students what they need to know to have a normal conversation. But again, it might not be the teachers fault the student ain't getting it...