1
2
![[fr]](/mwf/v12/flags/fr.png "France")
hello,
I look for the performance's difference ( kn's or percentage) between processor I5 750 (or same strenght) and I7 920 or I7950 for rybka 4. The I7 i very expensive for me and i want to evaluate the differences.
We say rybka 4 is very good for I7 with instructions but this specification is big ?
thank you
I look for the performance's difference ( kn's or percentage) between processor I5 750 (or same strenght) and I7 920 or I7950 for rybka 4. The I7 i very expensive for me and i want to evaluate the differences.
We say rybka 4 is very good for I7 with instructions but this specification is big ?
thank you
You didn't consider the i7 930!
http://www.bit-tech.net/hardware/cpus/2010/05/19/intel-core-i7-920-versus-i7-930/1
http://www.newegg.com/Product/Product.aspx?Item=N82E16819115225&cm_re=i7_930-_-19-115-225-_-Product
The price of this core is also decreasing and the i7 920 is being offered less as an option.
http://www.bit-tech.net/hardware/cpus/2010/05/19/intel-core-i7-920-versus-i7-930/1
http://www.newegg.com/Product/Product.aspx?Item=N82E16819115225&cm_re=i7_930-_-19-115-225-_-Product
The price of this core is also decreasing and the i7 920 is being offered less as an option.
![[nl]](/mwf/v12/flags/nl.png "Netherlands")
It is a lot cheaper to overclock the i7 930!
You're correct- I was looking at the 960!
![[za]](/mwf/v12/flags/za.png "South Africa")
The guy said i7 is very expensive for him
Yeah, I noticed this myself about the 950, cheaper than 920 or 930?
When we were talking about the 950 at the outset of the summer it was a little over $500.
![[gb]](/mwf/v12/flags/gb.png "United Kingdom")
http://sedatchess.110mb.com/index.php?p=1_13
These are Rybka 3 benchmarks. Also be aware that many of the benhcmarks are overclocked.
But this list will tell you pretty much what you need to know.
These are Rybka 3 benchmarks. Also be aware that many of the benhcmarks are overclocked.
But this list will tell you pretty much what you need to know.
i know sedatchess but it's rybka 3, overclocked as you say for most, and as i said rybka 4 is for i7 :
The difference between i5 et i7 with rybka 3 = difference between i5 and i 7 with rybka 4 ?? if yes then sedatchess is good for me.
The difference between i5 et i7 with rybka 3 = difference between i5 and i 7 with rybka 4 ?? if yes then sedatchess is good for me.
Dear zezette,
I'm also french and I'm not here to give you any advice on i7 or whatever. I don't understand n othing about that. Just to say than, as french, it's very funny to see your nicname. I guess that few foreign people understand the joke there is inside.
Respect
I'm also french and I'm not here to give you any advice on i7 or whatever. I don't understand n othing about that. Just to say than, as french, it's very funny to see your nicname. I guess that few foreign people understand the joke there is inside.
Respect
Save your money for a x980!
Hello maxou, thanks :-)) but my nicname it's for "the pere noel est une ordure" !!!
I'm just finished to study test rybka 3 : i would like to specify i want to keep my computer 3 or 4 years and WITHOUT overclocking (perhaps it's strange but i afraid to burn something, i don't master).
In this test, the i5 750 is better i7 920 and i7 940 without overclocking!!! why ??
And for the fritz 9 chess benchmark, i7 920 is better i5 750.
test with Same processor, results different for rybka and fritz (between the 2), it's possible ???
For rybka 3, i see that Phenom II x4 965 is the top cheap, but no test for fritz (it's the x6)
it's difficult....
In this test, the i5 750 is better i7 920 and i7 940 without overclocking!!! why ??
And for the fritz 9 chess benchmark, i7 920 is better i5 750.
test with Same processor, results different for rybka and fritz (between the 2), it's possible ???
For rybka 3, i see that Phenom II x4 965 is the top cheap, but no test for fritz (it's the x6)
it's difficult....
Dear zezette,
that's also what I thought about because this character is very famous in France. On the other hand, may I ask you what GUI do you use to make Rybka running ?
Best regards
that's also what I thought about because this character is very famous in France. On the other hand, may I ask you what GUI do you use to make Rybka running ?
Best regards
i use chessbase !
same thing!
I use Aquarium
so do I!
Yeah, I'm thinking the 980x will stay high for awhile just like the QX6700 and QX6800 have done. It might be another 1 1/2 yrs before they cut these prices in 1/2. and by then you will be able to get a nice entry level Ivy Bridge cpu that runs just as quick or quicker for 1/4 the price. But isn't that always the case with the advancement of technology! My M.O. is to buy a new PC when a current mid to high-level CPU is scoring 10 times faster than my old hardware or about every 5 years.
I was looking at something to replace my Core 2 Duo- and because the QX6800 is still going new for around $960 or more. I would have to be satisfied with up-grading the Formula Maximus to a Core 2 quad Q9550. I'd like to have that board functioning with something decent- I could wait for the QX6800 to cleave in price.
and save for the more expensive upgrade in the x980.
Right now I'm good with the 930- it suites my purposes. I'm not in any hurry to move up just yet.
I think the bottom line is to focus on having mobo's with some kind of viable upgrade options built in.
and save for the more expensive upgrade in the x980.
Right now I'm good with the 930- it suites my purposes. I'm not in any hurry to move up just yet.
I think the bottom line is to focus on having mobo's with some kind of viable upgrade options built in.
I would go for the QX6700 its about $500 cheaper also how is that 930 ovrclocking? you getting @ least 3.8 ghz on air?
I have had it overclocked at 4200 but recently I've kept it at 4009 with temps at about 56c - that's at a frequency of 190 with a Cool Mater V8- I'm now looking into water cooling- I know I can easily go up to 200.
http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?pid=284703#pid284703
that is with the Cool Master V8-
that is with the Cool Master V8-
really cool temps! I'm impressed and fast!
I think it is a reasonable assumption that the relative performance of R3 wil give a very strong indication of the relative performance of R4.
I do not have figures to back that up specifically. However, I can say that as I have changed PC's all my chess programs (R3, Shredder 4 and 9 and Hiarcs 7.32) have all speeded up fairly much in proportion, so I would expect R4 to do much the same.
I do not have figures to back that up specifically. However, I can say that as I have changed PC's all my chess programs (R3, Shredder 4 and 9 and Hiarcs 7.32) have all speeded up fairly much in proportion, so I would expect R4 to do much the same.
ok thank you regularisar, i 'm going to take the results to R3.
But for the results i5 750 better i7 920, what do you think ?
But for the results i5 750 better i7 920, what do you think ?
The i7 920 is being phased out! If you want one-you'd better act quick!
http://www.newegg.com/Product/Product.aspx?Item=N82E16819115202
http://www.newegg.com/Product/Product.aspx?Item=N82E16819115202
it's sold out!
Now I have a collectors item
Now I have a collectors item
The 950s are cheaper now!
For chess the 750 looks much better value to me.
The main advantages of the 920 are that:
- it has 3 memory controllers as oppsoed to the 750's 2 memory controllers, but this is of limited relveance to chess as chess is memory latency limited rather than memory bandwidth limited
- it supports hyperthreading qhwewas the 750 does not, but normally hyperthreading is not useful to chess programs. If you wanted to run 8 copies of Rybka each using a single thread then this would be relevant. If on the other hand you are happy with 4 or less copies of Rybka running then hyperthreading offers little advantage. If you just want maximum power throughonecopy of Rybka then hyperthreading is a disadvantage.
The 750 is just cheaper and for most people just about as good for chess as the 920.
The main advantages of the 920 are that:
- it has 3 memory controllers as oppsoed to the 750's 2 memory controllers, but this is of limited relveance to chess as chess is memory latency limited rather than memory bandwidth limited
- it supports hyperthreading qhwewas the 750 does not, but normally hyperthreading is not useful to chess programs. If you wanted to run 8 copies of Rybka each using a single thread then this would be relevant. If on the other hand you are happy with 4 or less copies of Rybka running then hyperthreading offers little advantage. If you just want maximum power throughonecopy of Rybka then hyperthreading is a disadvantage.
The 750 is just cheaper and for most people just about as good for chess as the 920.
i don't understood very well, the 4 copies or 8 copies but it's nothing...:-) i retain i7 950 it's better.
My last question, i 'm going to buy my computer today or tomorrow :
I retain two cores : i7 950 and Phenom II X6 1090 T , expensive but i keep it for long time!
i prefer intel just for reputation, i don't know AMD. 5 or 6 years ago, we said me temperature 's AMD was bof bof...
But for chess, if cores is very important, the best is Phenom. And for core 's speed too.
In passmark, the I7 is slightly better to Phenom : 6275 vs 6076.
Finally, it's perhaps just a matter of taste ?
coul you tell me your opinion please ?
thank you for your help!!
My last question, i 'm going to buy my computer today or tomorrow :
I retain two cores : i7 950 and Phenom II X6 1090 T , expensive but i keep it for long time!
i prefer intel just for reputation, i don't know AMD. 5 or 6 years ago, we said me temperature 's AMD was bof bof...
But for chess, if cores is very important, the best is Phenom. And for core 's speed too.
In passmark, the I7 is slightly better to Phenom : 6275 vs 6076.
Finally, it's perhaps just a matter of taste ?
coul you tell me your opinion please ?
thank you for your help!!
![[be]](/mwf/v12/flags/be.png "Belgium")
The cpu today for best Price/Chess performance then i go for the AMD X6 1090T !
You have real 6cores..even you overclock the 950 higher ,you still have only 4 cores!
And the X6 1090T is also not a bad overclocker cpu!
What count if you use the computer for chess is to get the highest nodes..then choose X6 1090T!
Kind regards,
JP.
You have real 6cores..even you overclock the 950 higher ,you still have only 4 cores!
And the X6 1090T is also not a bad overclocker cpu!
What count if you use the computer for chess is to get the highest nodes..then choose X6 1090T!
Kind regards,
JP.
![[us]](/mwf/v12/flags/us.png "United States")
It is best to wait. dramatically better CPUs will be here soon. AMD's Bulldozer is likely to be a killer especially the Interlagos. Intel also will have a big jump with Sandy Bridge. Each should be released soon.
The Interlagos is 16-cores. And it will have 64 integer pipelines (chess engines are integer based generally). Just cause Interlagos is a server chip does not mean you have to buy a board with 2 sockets. They should have single socket boards capable of using the chip. But 2-socket and 4-socket will be available. In 4-socket that would be 256 integer pipelines. That should play a sweet game of chess.
Sandy Bridge will probably be nice too. But the particular path AMD is taking looks very good theoretically for chess.
Phenom will probably disappoint. Currently, chess engines run better on Intel.
The Interlagos is 16-cores. And it will have 64 integer pipelines (chess engines are integer based generally). Just cause Interlagos is a server chip does not mean you have to buy a board with 2 sockets. They should have single socket boards capable of using the chip. But 2-socket and 4-socket will be available. In 4-socket that would be 256 integer pipelines. That should play a sweet game of chess.
Sandy Bridge will probably be nice too. But the particular path AMD is taking looks very good theoretically for chess.
Phenom will probably disappoint. Currently, chess engines run better on Intel.
Today, you say intel is better and tommorow probably amd is better ?
but if i wait next year, you can say same thing perhaps, and i never buy my computer
Jpqy said to get the highest nodes, why do you think intel i better ?
but if i wait next year, you can say same thing perhaps, and i never buy my computer
Jpqy said to get the highest nodes, why do you think intel i better ?
I am not sure if Intel says this every release, but they said that Sandy Bridge represents the largest improvement ever by Intel. Could be hype, but somehow I doubt it. And yes there will be a shrink of Sandy Bridge very quickly. That's kinda new, it usually takes a while to do the "Tock" of their "Tick-Tock" cycle but it is looking like it will only be 6 months later. I think this is to attempt to avoid the risk of being #2 for a while. Bulldozer could potentially blow Sandy Bridge away. It is anyone's guess if that will be the case. They are going to quickly move to 22nm (Ivy Bridge). Probably hoping the extra speed can overtake the Bulldozer. It may really clock crazy.
I know I bought a 486 and was kicking myself when Pentium came out two months later. If we are going to see that kind of jump, I sure would not want to buy now.
The AMD chip is an altogether different chip than we have ever seen. AMD has been using basically the same building block since around 2003. This is going to be a completely new module. And you can bet they are sick of being #2. It is a completely new module as I said and Interlogos has 64 integer pipelines and they also said it was going to have a faster clock. That is because they are finally going to move to 32nm which was part of what was killing them. More pipelines and going faster has to be a good thing. Is it going to go fast enough to beat Ivy Bridge? Probably not in non-mulithreaded apps but my guess is that Intel is going to have to add some cores to keep up with multithreaded apps.
I like AMD's approach in their new servers. Since they always seem to be behind Intel in process shrink, they simply made a massive socket. You don't have to be at the same scale if you just shove a lot of cores in a socket. Take a look at the monster: https://www.evga.com/forums/tm.aspx?m=270365&mpage=1&print=true
Here is an interesting article: http://www.anandtech.com/print/3863
The latest: http://blogs.pcmag.com/miller/2010/10/amd_shows_llano_intel_talks_22.php
I know I bought a 486 and was kicking myself when Pentium came out two months later. If we are going to see that kind of jump, I sure would not want to buy now.
The AMD chip is an altogether different chip than we have ever seen. AMD has been using basically the same building block since around 2003. This is going to be a completely new module. And you can bet they are sick of being #2. It is a completely new module as I said and Interlogos has 64 integer pipelines and they also said it was going to have a faster clock. That is because they are finally going to move to 32nm which was part of what was killing them. More pipelines and going faster has to be a good thing. Is it going to go fast enough to beat Ivy Bridge? Probably not in non-mulithreaded apps but my guess is that Intel is going to have to add some cores to keep up with multithreaded apps.
I like AMD's approach in their new servers. Since they always seem to be behind Intel in process shrink, they simply made a massive socket. You don't have to be at the same scale if you just shove a lot of cores in a socket. Take a look at the monster: https://www.evga.com/forums/tm.aspx?m=270365&mpage=1&print=true
Here is an interesting article: http://www.anandtech.com/print/3863
The latest: http://blogs.pcmag.com/miller/2010/10/amd_shows_llano_intel_talks_22.php
thanks for this, i'll wait for six months and spend some dough to blow Paul out of the water
anandtech ?!
anandtech ?!
> I think this is to attempt to avoid the risk of being #2 for a while...
I don't think there is any doubt that Intel is always well aware of what is going on in there research and development division. The cost falls squarely on the consumer if they want to be foolish enough to ride the waves with Intel and stay ahead of the pack. I think in this case it is wise to let these companies fight it out over supremcy and hold off on impulse buying.
But there will always be someone who thinks they need the latest and if they can afford to jump every time Intel proclaims a break through-more power to them. I'm waiting for the time when they muck things up in haste and dump some miscalculated mistake on to the shelves.
Intel has many big advantages over AMD, and AMD has mucked up a lot more releases than Intel. Betting on AMD to have the top engine platform is really betting on a long shot...
> Betting on AMD to have the top engine platform is really betting on a long shot...
I've always gone with Intel. If you are right then there is no real competition between the two and may be Intel will not reach a point where they are hasty in putting out a premature product. Still no reason to jump everytime Intel says it is time to-unless of course you have to bucks to jump along with them!
i think Alan bought a big slice of Intel shares,wisely in my opinion.
I have a jaundiced outlook towards the AMD/Intel debate because I can't help but wonder if they are unofficially, the same company.
No way! I don't do straddles!
Alan,
So! You go globe trotting for Intel!
So! You go globe trotting for Intel!
When the Athlons first came out they were 50%+ faster than Intel's stuff and then they were the first to go 64-bit. And as far as I know Intel is the only one that let a chip out the door that actually got an instruction wrong (someone who knows more can correct me if this is not accurate). AMD has had a dry spell but I think--at least for chess--we are going to see a nice improvement. Hopefully they will not drag their feet much longer and get the new processors out there so they can have a day in the sun rather than remaining in the shadow of Intel. If they are so slow as to release them after Intel launches Ivy Bridge, I don't know how they will survive. But it sounds like they will get it out soon.
If Ivy Bridge does not get close to Bulldozer Intel could be in trouble for more than a year. Conceivably they could loose a third of the market. There will probably be a year and a half+ after Ivy Bridge before Haswell possibly two years because the "tock" (Ivy Bridge) was accelerated to only 6 months after Sandy Bridge. I have a sneaky suspicion they will have two "tocks" if they get in trouble.
You also have to wonder about Intel's long term strategy. They stay ahead by being the first to use a smaller process. How long can that work? It can only get so small where each successive shrink gets more difficult and there are diminishing returns from increasing quantum effects.
I do think both companies targets of lowering energy is correct because when shrinking further is no longer an option they will have to do some serious 3-dimensional stacking and it will not be so easy to cool the chips at least without solid state cooling built into the chips like layering a nanotech version of a Peltier that pumps heat laterally instead. Peltiers have a long way to go to be viable...currently they are terribly inefficient.
If Ivy Bridge does not get close to Bulldozer Intel could be in trouble for more than a year. Conceivably they could loose a third of the market. There will probably be a year and a half+ after Ivy Bridge before Haswell possibly two years because the "tock" (Ivy Bridge) was accelerated to only 6 months after Sandy Bridge. I have a sneaky suspicion they will have two "tocks" if they get in trouble.
You also have to wonder about Intel's long term strategy. They stay ahead by being the first to use a smaller process. How long can that work? It can only get so small where each successive shrink gets more difficult and there are diminishing returns from increasing quantum effects.
I do think both companies targets of lowering energy is correct because when shrinking further is no longer an option they will have to do some serious 3-dimensional stacking and it will not be so easy to cool the chips at least without solid state cooling built into the chips like layering a nanotech version of a Peltier that pumps heat laterally instead. Peltiers have a long way to go to be viable...currently they are terribly inefficient.
You have no idea what the Kn/s (the measure of nodes) is going to be of the future products. I doubt any Phenom is going to match Intel 920 or higher at the moment. And while I was thinking the Magny-Cours would be a great choice others have splashed cold water on that idea. I have yet to see any concrete numbers yet though.
I thought something like this would be a nice little chess computer: http://www.newegg.com/Product/Product.aspx?Item=N82E16813182230&cm_re=g34-_-13-182-230-_-Product
http://www.newegg.com/Product/Product.aspx?Item=N82E16819105267&cm_re=g34-_-19-105-267-_-Product
And an overclock should be possible with software. The board itself will not do it. If water cooling, you will probably have to make your own water block.
That is 12 true cores.
Something like this with the Bulldozer (also uses the same socket and probably the same board with a BIOS update) will probably be a lot of bang for the buck.
I thought something like this would be a nice little chess computer: http://www.newegg.com/Product/Product.aspx?Item=N82E16813182230&cm_re=g34-_-13-182-230-_-Product
http://www.newegg.com/Product/Product.aspx?Item=N82E16819105267&cm_re=g34-_-19-105-267-_-Product
And an overclock should be possible with software. The board itself will not do it. If water cooling, you will probably have to make your own water block.
That is 12 true cores.
Something like this with the Bulldozer (also uses the same socket and probably the same board with a BIOS update) will probably be a lot of bang for the buck.
this is quite amazing, 48 cores for $7900 excl. memory
http://cgi.ebay.com/TYAN-TX46-QUAD-MAGNY-COURS-6172-G34-Socket-S8812-MB_W0QQitemZ270534466899QQcmdZViewItemQQimsxZ20100227?IMSfp=TL100227144010r2061
how many KN/s would this give with Rybka 4 from starting position?
http://cgi.ebay.com/TYAN-TX46-QUAD-MAGNY-COURS-6172-G34-Socket-S8812-MB_W0QQitemZ270534466899QQcmdZViewItemQQimsxZ20100227?IMSfp=TL100227144010r2061
how many KN/s would this give with Rybka 4 from starting position?
1
2
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill