Best Unit for the Job?
 

Changing my mind from a previous thread, I am just going to save up and buy a more powerful crunching unit instead of just a new CPU. So...my question is this. Without being prohibitively expensive, and just looking for a computer that will be used mainly for number-crunching, what is the best combination of CPU, RAM, Motherboard, etc for performing large LL tests 45M plus and eventually going for the hundred million digit challenge? Also, are there any CPUs coming out in the next year or two that are going to be breakthroughs so to speak on LL speed, etc? Thanks.

Also, what sites are the best for custom building such a machine?

[quote=Primeinator;178601]Also, what sites are the best for custom building such a machine?[/quote]
you home is the best site
it is often possible to save quite a bit of money

I would probably recommend an i7 920 currently for good bang per buck for one pc.
If you are willing to pay the electricity for several pcs i think it is possible to get 2-3 Core 2 Quads for the same money.

[QUOTE=Primeinator;178601] eventually going for the hundred million digit challenge? Also, are there any CPUs coming out in the next year or two that are going to be breakthroughs so to speak on LL speed, etc? Thanks.
[/QUOTE]

I suspect by the time we are seriously shooting for the 100 million digits prize, anything you could buy today, or even this year, will be completely out of date, probably consigned to test factoring.

[quote=lfm;178673]I suspect by the time we are seriously shooting for the 100 million digits prize, anything you could buy today, or even this year, will be completely out of date, probably consigned to test factoring.[/quote]

I don't think so. With a Core i7 920 (nice overclockable to 3,7 Ghz) and 8 hyperthreads you can decrease your time for a 100M number to around 3-7 years ...

[QUOTE=joblack;178686]I don't think so. With a Core i7 920 (nice overclockable to 3,7 Ghz) and 8 hyperthreads you can decrease your time for a 100M number to around 3-7 years ...[/QUOTE]

OK, consider the machines that take 3-7 years to do today's exponents. (33-45 million bits) What would you recommend people do with them today? Do you think those machines had any reasonable chance of finding the first 10 million digit prime? Seems pretty slim odds to me.

[quote=lfm;178690]OK, consider the machines that take 3-7 years to do today's exponents. (33-45 million bits) What would you recommend people do with them today? Do you think those machines had any reasonable chance of finding the first 10 million digit prime? Seems pretty slim odds to me.[/quote]

You mean the first 100M digits prime? Perhaps its only 2 years (it's hard to estimate the time of the Core i7 if you don't have one) - the chance to find a 10M prime numbers isn't that much higher.

Anyway with a Core i7 I would definitely go for a 100M number (as I started some weeks ago -> you need to make regular backups including a RAID1). If the machine is outdated you can update it to a higher one (then you can still use this one for office work and smaller exponents). With Murphy's Law you'll double processing power every 2 years so you could upgrade then ...

The important step is to start calculating and perhaps upgrade to a higher cpu in the future ...

[quote=joblack;178699]...With Murphy's Law you'll double processing power every 2 years so you could upgrade then ...[/quote]
I think Murphy's Law is something else. :wink:

[quote=Flatlander;178701]I think Murphy's Law is something else. :wink:[/quote]

LMAO my fault - of course I mean Moore's Law ;).

I think Murphy's Law does apply here. It states that the box will be blown by lightning at 99% of a 100M digit test. (Could actually be more likely than the result being prime.) :rolleyes:

[QUOTE=Flatlander;178706]I think Murphy's Law does apply here. It states that the box will be blown by lightning at 99% of a 100M digit test. (Could actually be more likely than the result being prime.) :rolleyes:[/QUOTE]

Which is why he said to make back ups and preferably store them on a USB drive or other device :smile:

Interesting though. I found [URL="http://www.tigerdirect.com/applications/SearchTools/item-details.asp?EdpNo=4826473&CatId=333"]this[/URL] system, which seems to be a very good price. What do you guys think, would it be good for quickly completing large LLs 45M +? I know it is not an i7...but it is still a fast quad core.

[quote=Primeinator;178601]Changing my mind from a previous thread, I am just going to save up and buy a more powerful crunching unit instead of just a new CPU. So...my question is this. Without being prohibitively expensive, and just looking for a computer that will be used mainly for number-crunching, what is the best combination of CPU, RAM, Motherboard, etc for performing large LL tests 45M plus and eventually going for the hundred million digit challenge? Also, are there any CPUs coming out in the next year or two that are going to be breakthroughs so to speak on LL speed, etc? Thanks.

Also, what sites are the best for custom building such a machine?[/quote]

Hmm break this down into 2 parts::

1) Components - intel core i7/nehalem (the 920 is nice so are some of the more expensive parts which should drop again in price this fall)
6 gigs of ddr3 dram - preferrably a high speed/low (CAS) latency
a 10,000 rpm fast hard drive [if you really want to spend a few more $$ a 15,000 rpm SAS]
blah blah psu+moboard+video ; would recommend a nicely vented case [& an extra cooling fan (or if you even want to go that extra mile some sort of add-on water cooling apparatus)]

Such a machine comes in under $650 - expect just under $700 with a 300 gig velociraptor ... expect more if you're going to config with an SAS drive [I'd think you'd be better off with a server board+box if you're going the SAS drive route though]

2) I tend to watch the appropriate forums & a few respectable component websites for very nice sales and put my family+friend+personal machines together. The only caveat you have to watch for is some of the return times given for defective products if you do not purchase your system components all at once; tis still very doable by just being sure you can test the components you do purchase. I may buy some extra memory when it's on sale - pop it into existing 4core moboards or a drive etc - if it works - kewl it can sit until I get the other components, if it fails - RMA back and wait for the replacement part(s) to test them.

Hmm not sure if i'll get brow-beaten for including any site links so I'll just mention a few I frequent weekly: overclockers forums (ocforums / cyber deals), hardforum (hot deals), newegg, frys & microcenter have a lot of "b&m" (brick and mortar) only deals besides their online sales.

[quote=joblack;178699]With Murphy's Law you'll do[/quote]all that [i]and still not find a Mersenne prime[/i].

[quote=Flatlander;178706]I think Murphy's Law does apply here. It states that the box will be blown by lightning at 99% of a 100M digit test. (Could actually be more likely than the result being prime.) :rolleyes:[/quote]

In that exponent range I would at least make daily backups (if not building a RAID system). Losing your data at 99% would be really annoying.

Is a super-fast hard drive really that necessary for prime hunting though? DDR3 makes sense...faster and better RAM. How good was the system I posted a link to?

[quote=cheesehead;178724]all that [I]and still not find a Mersenne prime[/I].[/quote]

If you would like to use Prime95 in a realistic manner you wouldn't use it anyway. Don't be that pessimistic :D ...

[quote=Primeinator;178736]Is a super-fast hard drive really that necessary for prime hunting though? DDR3 makes sense...faster and better RAM. How good was the system I posted a link to?[/quote]

No it isn't ... a cheaper ASRock board is enough. You also don't need more than 4 Gigs RAM but a 64 Bit OS is better (gives you a 10 percent speed advantage because of the additional SSE registers). Think about creating a RAID1 (included in the board - just use two harddisks) and an additional USB Stick for backup purposes. For overclocking don't go over 3,7 Ghz (for the 920) because over that the Watt/Performance ratio decreases (check tomshardware.com for more overclocking information) ...

[QUOTE=joblack;178737]If you would like to use Prime95 in a realistic manner you wouldn't use it anyway. Don't be that pessimistic :D ...[/QUOTE]

I think that was just poking fun at your (mistaken) invocation of Murphy's Law.

[QUOTE=joblack;178699][QUOTE=lfm;178690]OK, consider the machines that take 3-7 years to do today's exponents. (33-45 million bits) What would you recommend people do with them today? Do you think those machines had any reasonable chance of finding the first 10 million digit prime? Seems pretty slim odds to me.[/QUOTE]You mean the first 100M digits prime? Perhaps its only 2 years (it's hard to estimate the time of the Core i7 if you don't have one) - the chance to find a 10M prime numbers isn't that much higher.[/QUOTE]No, read what he said again.

If you want the best desktop PC, get an i7 975 with 12 GB DDR3 2000 MHz RAM and overclock it.

If you want a great desktop PC get an i7 920 (make sure it's one with the D0 stepping) with 3 GB DDR3 1333 MHz RAM, or better yet 1600 MHz RAM. Overclocking would net you more performance, but you'd need an aftermarket cooler and you SHOULD NOT attempt this without research, you can easily burn out the motherboard or CPU if you do not know what you are doing. I'd say most people who overclock don't kill their equipment more out of luck than skill.

If you want a cheap but still good desktop PC, get a low end 45nm Core2Quad with 2 GB DDR2 1066 MHz RAM, there should be some overclocking headroom even with the default cooler and they are easier to OC than the i7s.

If you want the best performance per watt, I'd say go with the above Core2Quad system, and downclock it.

The Phenom and Phenom II series CPUs from AMD will offer decent performance too, but the highest end Phenom II CPU still only performs about on par with the lowest i7.

[QUOTE=lavalamp;178751]The Phenom and Phenom II series CPUs from AMD will offer decent performance too, but the highest end Phenom II CPU still only performs about on par with the lowest i7.[/QUOTE]

The Phenom II is a great chip -- I'm running one right now. But the original Phenom isn't worth buying, especially with how cheap a Phenom II is:[LIST][*] [URL="http://www.newegg.com/Product/Product.aspx?Item=N82E16819103674"]$190[/URL] for 3.0 GHz[*] [URL="http://www.newegg.com/Product/Product.aspx?Item=N82E16819103650"]$140[/URL] for a 2.6 GHz[/LIST]And don't forget that motherboards for the Phenom II are cheaper than mobos for the i7 (save $80 to $150, assuming you use an AM2+ rather than AM3) and that you can use DDR2 instead of DDR3 (save $100).

Overall a Phenom II system can be $320 to $350+ cheaper than a similarly-clocked i7-based system. (Of course Prime95 is better optimized for the i7, so if that's what you're doing it's better to compare to a Phenom II running at a higher speed.)

But a Phenom system will save only about $50 over the Phenom II, and its actual performance is much lower (cache issues, Cool n' Quiet problems, etc.).

[QUOTE]Also, what sites are the best for custom building such a machine?[/QUOTE]
Of course I would not spent money just to run mprime faster.

But for example if you are interested in some kind of hosting webservices anyway, then I would rent such fast machine.
In Germany its very cheap, here for example:
[url]http://www.hetzner.de/de/hosting/produktmatrix/rootserver-produktmatrix/[/url]
Intel® Core™ i7-920
8GB RAM
49 EUR /month
IMO very cheap - just running such thing 24/7 at home would cost you ~20 EUR electricity per month plus buying the hardware.
And there you have additionally a 100 Mbit connection, traffic unlimited to use it for a lot of other things.
Also if it produces a Bad LL - they will give you a new machine within some hours.
On next cpu generation you can just switch to the next special offer.

So a quad Phemon II at 3.0 Ghz would be a good choice?

How good is this system? [url]http://www.tigerdirect.com/applications/SearchTools/item-details.asp?EdpNo=4826473&CatId=333[/url]

[QUOTE=Primeinator;178601]Changing my mind from a previous thread, I am just going to save up and buy a more powerful crunching unit instead of just a new CPU. So...my question is this. Without being prohibitively expensive, and just looking for a computer that will be used mainly for number-crunching, what is the best combination of CPU, RAM, Motherboard, etc for performing large LL tests 45M plus and eventually going for the hundred million digit challenge? Also, are there any CPUs coming out in the next year or two that are going to be breakthroughs so to speak on LL speed, etc? Thanks.

Also, what sites are the best for custom building such a machine?[/QUOTE]

The question, as asked, is poorly conceived. The following
will be an intro to economic decision analysis. The question
you ask can be modelled as a linear program.

(1) What are the constraints? I assume you have a constraint
(i.e. upper bound) on cost. Do you consider just the purchase
price or do you include operating costs? Do you have a constraint
on your time? Managing multiple systems as opposed to a single
system will require more time. Multiple systems will have greater
operating costs. It might be cheaper to (say) buy 4 lesser systems
(which run at say 1/3 to 1/2 the speed) or a single very fast
system. The former will have greater throughput, the latter
lower latency.

(2) Maximizing "bang for the buck" is meaningless. You can
always get more performance by spending more. OTOH, you can
probably find a 10-year old used system for $5.00, which will
be 50 times slower than a new system. But at only $5.00 you
would be getting greater "bang for the buck" (i.e. price
performance)

(3) Are you looking to maximize price/performance? (with say a
lower bound on performance?)
Are you looking to maximize throughput?
Are you looking to minimize latency (time to test a single
number)?

These are not the same question.


Specify your variables. Specify your constraints. Specify your objective.

Variables will be a measure of perfomance for each system and the cost of
each system. Another variable might be the electrical costs.
Constraints will be a lower bound on performance, perhaps a bound on
price/performance and an upper bound on costs. The objective is up
to you.

Solve.
Then do a sensitivity analysis. Compute your shadow prices. Might
you be willing to relax your upper bound on cost by e.g. 10%, if
it means a 20% improvement? etc. etc.


FINALLY: You can always get a better system for the same price by
waiting e.g. a year. This should never be a consideration unless your
objective is to maximize throughput over a *fixed time period* starting
today. i.e. if you want to test the most numbers over the next 5 years
it might pay to wait.

Thank you for elucidating the inadequacy of my verbiage in my original question. I should clarify a few things. First, electricity cost. For the foreseeable future, the cost of electricity will be included in my residency meaning I will not have to pay more if I hook up a powerful machine. That being said, questions are going to be asked if I somehow place a server in my room that places a significant jump in energy demands.

As for using several less powerful and expensive machines as opposed to one more powerful machine, I do not necessarily have the room to operate a farm. Thus, one, possibly two machines (in addition to what I have running) is what I'm considering, but leaning more towards the one machine option.

For the cost of the machine, I do not have a fortune to spend. Preferably less than $700 USA dollars. However, if something could offer a significant jump in performance for little additional relative cost, then I would most definitely consider that option. As for waiting for new technology, my current focus is to find a new, world-record prime with a long-term goal of finding a 100M digit prime. Unless some new technology is planned to be released in the coming few years that promises a significant increase in LL speed, I would like to go ahead and get a machine now and in the future get a machine capable of testing a 100M candidate in a reasonable time (when such technology exists). The planned Intel Nehalem EX seems like a powerful candidate-- 8 cores at 16 threads per core w/ DDR4 (quad-channel) memory. However, such a processor is most likely going to be very expensive and is probably not meant for desktop use.

[quote=R.D. Silverman;178761](2) Maximizing "bang for the buck" is meaningless. You can always get more performance by spending more. OTOH, you can probably find a 10-year old used system for $5.00, which will be 50 times slower than a new system. But at only $5.00 you would be getting greater "bang for the buck" (i.e. price performance)[/quote]Your final sentence seemingly contradicts your opening sentence.

Maximizing "bang for the buck" is just what linear programming [U]is[/U] good for.

Now, if you want to point out that Primeinator is not actually maximizing "bang for the buck" even though he's stated that as a goal -- IOW, he's not doing the linear programming correctly -- which is what you point out later -- that's okay.

So to (begin to) satisfy the linear program proponents, let's write this one out a bit more.

It looks like you want to maximize Prime95 throughput over a span of, say, two years for a total cost of $700 not including electricity, subject to the constraint of "not more than two computers at a time". Fair?

Goddamn it, that reminds me of the Simplex Method. Doing that by hand was horrible.

[QUOTE=Primeinator;178768]Thank you for elucidating the inadequacy of my verbiage in my original question. I should clarify a few things. First, electricity cost. For the foreseeable future, the cost of electricity will be included in my residency meaning I will not have to pay more if I hook up a powerful machine. That being said, questions are going to be asked if I somehow place a server in my room that places a significant jump in energy demands.

As for using several less powerful and expensive machines as opposed to one more powerful machine, I do not necessarily have the room to operate a farm. Thus, one, possibly two machines (in addition to what I have running) is what I'm considering, but leaning more towards the one machine option.

For the cost of the machine, I do not have a fortune to spend. Preferably less than $700 USA dollars. However, if something could offer a significant jump in performance for little additional relative cost, then I would most definitely consider that option. As for waiting for new technology, my current focus is to find a new, world-record prime with a long-term goal of finding a 100M digit prime. Unless some new technology is planned to be released in the coming few years that promises a significant increase in LL speed, I would like to go ahead and get a machine now and in the future get a machine capable of testing a 100M candidate in a reasonable time (when such technology exists). The planned Intel Nehalem EX seems like a powerful candidate-- 8 cores at 16 threads per core w/ DDR4 (quad-channel) memory. However, such a processor is most likely going to be very expensive and is probably not meant for desktop use.[/QUOTE]

The Nehalem-EX aka Beckton, will be a server CPU designed for use in quad CPU servers (read: extremely expensive and you'll need 4), it will have 8 cores, and [color=red]2 threads per core[/color] for a total of 16 threads, so 64 threads for a quad CPU system. It will also use [color=red]DDR3[/color] memory which will be fully buffered (increasing the access latency and cost, but generally running at a lower frequency).

Put it this way, you will not be buying a fully kitted out Beckton system any time soon.

The next high end desktop CPU from Intel will be the Westmere with 6 cores (so 12 threads total), expect it to make a showing towards the end of this year or beginning of next year. Then after that it will be Sandy Bridge which will come in 4 and 8 core varients and have twice the performance per clock per core and run at a higher default frequency, expect that about a year after Westmere.

Do I assume you are in the US then? You didn't actually mention where you were when you asked for web sites.

[quote=lavalamp;178791]The Nehalem-EX aka Beckton, will be a server CPU designed for use in quad CPU servers (read: extremely expensive and you'll need 4), it will have 8 cores, and [COLOR=red]2 threads per core[/COLOR] for a total of 16 threads, so 64 threads for a quad CPU system. It will also use [COLOR=red]DDR3[/COLOR] memory which will be fully buffered (increasing the access latency and cost, but generally running at a lower frequency).[/quote]

The Nehalem EX will be a nice processor but the price with be also impressive - you won't get the machine for $700 ... ;)

The system you linked to on TigerDirect will be a decently performing quad core system, though personally I'd like faster RAM in it. You could have two of those for your $700 budget and together they would outpace a single i7 system. They would use more power though obviously, which means it'd cost more to run and heat your place up more.

Here's how you could get a good i7 system for $700:[code]i7 920 & OCZ Reaper HPC 3GB (3 x 1GB) 240-Pin DDR3 SDRAM DDR3 1866 (PC3 15000)
http://www.newegg.com/Product/ComboDealDetails.aspx?ItemList=Combo.202682
$333.98 (- \10 rebate)


GIGABYTE GA-EX58-UD3R
http://www.newegg.com/Product/Product.aspx?Item=N82E16813128375
$199.99 (- $15 rebate)


Recertified: Seagate Barracuda 7200.8 400GB
http://www.newegg.com/Product/Product.aspx?Item=N82E16822148432
$39.99


ASUS EAH3450/DI/256M Radeon HD 3450 256MB
http://www.newegg.com/Product/Product.aspx?Item=N82E16814121259
$30.99 (- $10 rebate)


Rosewill R222-P-BK Black Steel ATX Mid Tower Computer Case
http://www.newegg.com/Product/Product.aspx?Item=N82E16811147095
$19.99


RAIDMAX RX-400AF 400W ATX12V / EPS12V 80 PLUS BRONZE
http://www.newegg.com/Product/Product.aspx?Item=N82E16817152039
$49.99 (- $20 rebate)


LITE-ON Black 18X DVD-ROM
http://www.newegg.com/Product/Product.aspx?Item=N82E16827106274
$17.99


$692.92 (- $55 rebates)[/code]If you wanted to try your hand at overclocking, an aftermarket cooler is practically a requirement. Unfortunately all the better aftermarket i7 coolers seem to be arranged in such a way as to not take advantage of the vent and duct over the CPU in the case I picked, but you could still take the duct off and fit a side fan there if you wanted. It seems the best cooler on Newegg for the i7 is [URL="http://www.newegg.com/Product/Product.aspx?Item=N82E16835608007"]this one[/URL], which tacks on another $74.99, but $5 can be shaved off by changing the bundles slightly. So the final price would be $762.92 with $55 rebates, assuming they go through OK, the total price minus delivery and tax would be $707.92.

Edit: I didn't include an operating system by the way, I'll leave you to decide that. Whatever you go for though, I'd recommend getting a 64bit version.

[quote=lavalamp;178793]The system you linked to on TigerDirect will be a decently performing quad core system, though personally I'd like faster RAM in it. You could have two of those for your $700 budget and together they would outpace a single i7 system. They would use more power though obviously, which means it'd cost more to run and heat your place up more.

Here's how you could get a good i7 system for $700:[code]i7 920 & OCZ Reaper HPC 3GB (3 x 1GB) 240-Pin DDR3 SDRAM DDR3 1866 (PC3 15000)
[URL]http://www.newegg.com/Product/ComboDealDetails.aspx?ItemList=Combo.202682[/URL]
$333.98 (- $10 rebate)[/code]
[/quote]

The number of ram sticks should always be even because of the memory interleave. Odd numbers decrease the performance.

[quote=joblack;178798]The number of ram sticks should always be even because of the memory interleave. Odd numbers decrease the performance.[/quote]

That is wrong for the i7. Helpful link (in German): [URL]http://www.pcgameshardware.de/aid,677383/Fuenf-Triple-Channel-Speicher-Kits-fuer-Core-i7-CPUs-im-Test-Vorschau-auf-PCGH-04/2009/RAM/Test/[/URL]

[quote=ckdo;178799]That is wrong for the i7. Helpful link (in German): [URL]http://www.pcgameshardware.de/aid,677383/Fuenf-Triple-Channel-Speicher-Kits-fuer-Core-i7-CPUs-im-Test-Vorschau-auf-PCGH-04/2009/RAM/Test/[/URL][/quote]

You're right, but that's obviously only true for the i7 - all other CPUs need even ram stick numbers ...

[QUOTE]Edit: I didn't include an operating system by the way, I'll leave you to decide that. Whatever you go for though, I'd recommend getting a 64bit version.[/QUOTE]

Linux 64-bit is what I'm thinking. I've heard Prime95 is faster on Linux anyway.

[QUOTE]Do I assume you are in the US then? You didn't actually mention where you were when you asked for web sites. [/QUOTE]

Yes, sorry. I should have indicated where I lived. I'm in the central United States.

So how much performance difference is there between the i7 920 and Q6700? Is the 6700 a 64-bit compatible CPU?

[quote=Primeinator;178811]Linux 64-bit is what I'm thinking. I've heard Prime95 is faster on Linux anyway.



Yes, sorry. I should have indicated where I lived. I'm in the central United States.

So how much performance difference is there between the i7 920 and Q6700? Is the 6700 a 64-bit compatible CPU?[/quote]

The core i7s see an "appreciable" performance boost on multi-threaded apps.

It is difficult to quantify in general but let's just toss out a 'sane range' of 20-50% performance boost for an i7 with fast ddr3 dram over a Q6700 with fast ddr2 dram on a typical cpu bound multi-threaded app. As someone has earlier pointed out the D0 steppings of the 920 apparently overclock very well. [ you can go read up on coolers & overclocking here - [URL]http://forums.extremeoverclocking.com/forumdisplay.php?f=84[/URL] ]

It goes without mention that some of the performance boost in going with a core i7 comes from memory interface aspects - i'm pretty sure unless i have overlooked some niche quad core moboard that slots the Q6xxx series chips & their pinouts that they all had a FSB. The moboard memory interface for the core i7s utilize Quick Path Interconnect.

The core i7 is a superior performer over phenom II for your intended use at this point in time. The AMD istanbuls with 6 cores are on par with the multithreaded nehalems ... but are more costly right now than the 920 [and who is to say that with the short time from 1st tape out to mass production that there are not too many undiscovered bugs in the istanbuls]

C++ya,
cout << "xkey" << endl;

I suggest to get the price worthy choice - means a Qxxxx quad core.

The Core i7 is a only a choice for ultra rich people with no other problems in life.

[QUOTE=joblack;179963]The Core i7 is a only a choice for ultra rich people with no other problems in life.[/QUOTE]

Not really. The Qxxxx series you suggest ranges from $160 to $350 on Newegg. The i7 920 is $270. Yes, the motherboards are more expensive, but they're not *that* bad.

Don't forget that the only desktop Nehalem CPUs out now are the Bloomfield high end i7s. Soon there will be mid-range 45nm Lynnfield and low-end 32nm Clarkdale chips on socket LGA-1156.

As I understand it, for that socket there will be i5s (at 45nm) released in 2 or 3 months, i3s (at 32nm) released in 5 or 6 months and i7s (at 45nm) released next year, they will still take DDR3 RAM, but won't use QPI and have only dual channel memory.

Here's a quick key to show what the number after the i means:

i9 = six cores, 12 threads (not yet released, but will be for the high end and servers)
i7 = four cores, eight threads
i5 = four cores, four threads (no hyperthreading)
i3 = two cores, four threads

[quote=lavalamp;179967] but won't use QPI and have only dual channel memory.[/quote]
which wont?
is it all the new i3s, i5s, i7s and i9s?
i would be surprised if the i9s didnt use QPI as it looks like a step back unless there is something new
edit: do you have any good links hand for finding this sort of thing?
googling has never worked very well for me for this type of subject

CPUs released on socket LGA-1156 will not use QPI. Those i3, i5 and i7 chips I mentioned in the coming months are for socket LGA-1156. Since the i9 is for high end, and LGA-1156 is distinctly not high end, the i9 chips will be on socket LGA-1366 for desktops and will have QPI.

As for finding stuff out, I just read a lot of tech sites and occasional press releases. If you're interested in tech then checkout Toms Hardware, [H]ard|OCP and AnandTech and basically anyone they link to.

AMD 12 Core coming soon....
 

[url]http://www.itworldcanada.com/Pages/Docbase/ViewArticle.aspx?id=idgml-9cdcd235-9c0a-45cd&Portal=4fb7319b-aa7c-423a-822d-2f6e24698c71&sub=490495[/url]

[QUOTE]AMD plans to launch later this quarter, an ultra low power Istanbul version of its Opteron EE offering. And, AMD has also said its 12-core Magny-Cours processor is slated for a 2010 release. [/QUOTE]

[quote=CRGreathouse;179964]Not really. The Qxxxx series you suggest ranges from $160 to $350 on Newegg. The i7 920 is $270. Yes, the motherboards are more expensive, but they're not *that* bad.[/quote]

You can get a quite good mainboard around 36 Euro (VGA included) - 2 x 2 GByte DDR2 - 800 around 38 Euro. The mainboards for the Core i7 are at least around 200 Euro and the memory (DDR3) is more expensive.

The question is if you want to have the most performance for GIMPS or for yourself (for instance you want to play hightech games, ...) - they aren't mutual exclusive so it's a matter of personal taste.

For the first I suggest get two or three of the E5400 (overclock them) + Mainboards and you have much more power for the same price ...

The i7 motherboards should cost well less than 200 euro, or you're getting ripped off. I could get them for less the last time (maybe 4-6 months ago) I was looking, and prices have surely fallen since.

But I don't disagree with your analysis. I chose a Phenom II rather than an i7 because of the cost of the motherboard.

[quote=CRGreathouse;180871]The i7 motherboards should cost well less than 200 euro, or you're getting ripped off. I could get them for less the last time (maybe 4-6 months ago) I was looking, and prices have surely fallen since.

But I don't disagree with your analysis. I chose a Phenom II rather than an i7 because of the cost of the motherboard.[/quote]

You're right - the price got down but is still around twice as much as an expensive Qxxx mainboard. The earlier Phenom had some problems with the TLB, today it might have more fun.

The fun thing for a i7 (Nehalem) is the the Hyperthreading and would be my first choice to hunt 100M primes ... but only for that reason ... ;).

[QUOTE=joblack;180883]The earlier Phenom had some problems with the TLB, today it might have more fun.[/QUOTE]

The Phenom was terrible. The Phenom II is great -- better than I expected.

[quote=CRGreathouse;180871]The i7 motherboards should cost well less than 200 euro, or you're getting ripped off. I could get them for less the last time (maybe 4-6 months ago) I was looking, and prices have surely fallen since.

But I don't disagree with your analysis. I chose a Phenom II rather than an i7 because of the cost of the motherboard.[/quote]

You're right - the price got down but is still around twice as much as an expensive Qxxx mainboard. The earlier Phenom had some problems with the TLB, today it might have more fun.

The fun thing for a i7 (Nehalem) is the the Hyperthreading and would be my first choice to hunt 100M primes ... but only for that reason ... ;).