nehalem design will force power users to use extreme cpu's - Page 2

Will that be done by Motorola-china?

Advertisement?

You'd be surprised. I have 10-year-old equipment and keep it on 24/7. Heck, the old Celeron 900 unit I got from a trash pile and turned into an HTPC has been running ever since I put its current HDD in it, for a 39 day uptime so far. My desktop runs 2/47 as well as it is a local file and print server and folds as well. It also takes about three minutes to boot and doesn't hibernate or suspend well due to the RAID card I have in it, so I leave it on. Its HDD SMART logs show they've been on for 21,608 hours and the PC was built in late February of 2006.

There are lots of reasons why somebody would have a computer on 24/7 and keep one for 10 years. HTPCs need to be on 24/7 to record your shows. The torrent box some people here have would be on 24/7 as well. Almost any kind of home server is going to be up 24/7 as well as it's pretty much a set it and forget it kind of affair. Do you really want to have to go and turn on the file server every time you want to back up your computer? No, you'll just leave it on. Old computers are useful as servers and standard-def HTPCs as well as being a Web kiosk type of machine. A 10-year-old PII or K6 machine will run a Web browser just fine to do things like look up recipes in the kitchen. If you spill a gallon of milk on it and kill it, big deal. Go pick another one off a curb somewhere. If you do the same with a new $1000 laptop, you'd be pissed.

intel just came out with a 3.5ghz quad core xenon cpu for $1500

this is simply a souped up, feature added, hand picked or machine cpu that is not much difference then a q6600 - my point? overclocking is simply taking advantage of circumstances.

that circumstance is that intel must milk a cpu design for years - the first quad is not much different from the last - intel slow turns up the speed to keep prices up.

so when you take your q6600 and turn it to 3.6ghz remember you just got a $1500 cpu!

I suspect that the likelihood of processors failing on their own is still pretty low even under overclocking conditions. Looking at it from the point of view of a processor manufacturer however, releasing processors with higher clock speeds (and therefore making them much more power hungry) still increases the risk even if it might be indirectly. As an example, look at the fiasco AMD had recently with people putting their high end processors (9850BE) on low end motherboards (780G).

Having enthusiasts trying to pair the low end motherboards with the high end processors that the motherboards weren't able to handle resulted in a litany of motherboard failures and likely quite a few processor failures as well. All of those processors almost certainly sent back to the manufacturer as defective despite the fact that the cause of their demise was a bunch of wingnuts who were either too ignorant or too lazy to read the supported processor list for their motherboards. Nevertheless, this results in some financial losses for the manufacturer.

Since the CPU manufacturers cannot control what kinds of motherboards their end users will pair with their processors, the best way for them to minimize the risk of this type of thing happening is to try to manufacture all of their processors to fit within certain power envelopes. Doing this increases the likelihood that the motherboards that will eventually be paired with them will be able to support the required load.

That's my take on the issue at any rate. If anyone else has some comments this might actually make an interesting discussion topic in and of itself.

I've never had a modern processor fail on me, even when the motherboard failed. I had one failure in the 386SX days, but the CPU was soldered on the motherboard and when it died, there was no way to troubleshoot, so I just bought a new one. Once in the my early transition to 486 and LIF, I didn't identify pin 1 and fried a 486DX2-50, but no such problems since the introduction of zif sockets.

AMD's fiasco was caused by marketing plus board partners cutting corners. The chipset has no problems with the higher clocked CPU's but most 780G boards weren't designed by partners with circuitry that could handle the extra watts:

http://www.anandtech.com/mb/showdo [...] i=3279&p=2

Follow up article here:

http://www.anandtech.com/mb/showdo [...] =3299&cp=2

I still wouldn't put a 125 watt processor on a 780G, but the perception that it's just a board for HTPC, SOHO and budget gaming is a bit offbase, as it can do well with 95 watt triple and quads. Hopefully, Gigabyte and ASUS will provide support for lower clocked 95 watt Denebs (which are rumored to arrive at 2.4 and 2.6 -- still better than most current B3's at stock).

If I ever choose to overclock, it wouldn't be on a 780G, but it's a great board at a great price when purchased from better partners who regularly provide new CPU support via bios updates.

Are you on drugs? So much incoherent rambling from every post you make.

You might also want to keep in mind that overclocking usually only exist in enthusiast community. Joe is not going to overclock his CPU, either because he doesn't know how, or he's not ready to do so. Data centers and server operators are also not likely to overclock their CPUs, because stability is what they want.

So while for most of us, we can easily overclock a CPU to compete against a much more expensive part, others don't mind shelling out the money for the peace of mind.

actually dude i never been warned since i do not advertise - but what ever.....


my point is that if intel is selling a 3.5ghz quad xeon that is not really much different then a q6600, and i posted 13 months ago the sweet spot is 3.6ghz for the g0 q6600 your wasting potential, you are missing on free performace gains by not overclocking.

just like if you could turn up your 200 hp motor to 325 hp with little change in gas mileage

its not an ad its my point, i have been doing this for years selling computers at or near what people say is too high voltage too fast speed and not one has come back.

i must agree sometimes i write these things while i am gaming or doing other things or its very late - what ever........

this is my little pet project for years and people still do not understand that overclocking is not going to hurt there computers - i ship systems at 1.5v all the time and never get one back not one! not one has failed - i get pm's and posts from the so called experts here who tell me not to tell people to do that - omg they will blow up the world!

oc is not cern!


and if i get banned i either get unbanned or i find a new spot to hang out - a few losers like to harasse me since i am not going to waste my time and write a blog - i am not sitting at my dead end desk job like many ...... i would guess thats why that stick is in your chair! lol!

anyways!!!!!!

find one person on THG that a sold a comptuer too, find one person that i pm .... find one post where i offered one other then used ones on classifieds - so get a life and do not post unless you have something to contribute to the title or too those acutally seek info!


if your not overlocking its because you do not understand the potential (unless its a laptop) - my post is pointed to power users -- 3d guys, traders and gamers those are the people who i sell systems too.

never ending thred

i was born 1979...yet i joined tomshardware Joined : Tue Jan 01, 1970

hehehehe...I just want nehalem to be affordable....i love to build something similar to AMD 780G + AMD 4850e combo

Really? Maybe we should talk to turpit, and see what he said.

Poor analogy here. First of all, like I said earlier, not everyone is willing to overclock. Overclocking usually requires substantial knowledge in computer architecture, and painstakingly long time in obtaining stability, as well as the highest overclock possible with lowest overvolt.

Now, no one here has ever questioned your ability to overclock. But just because you know how to overclock doesn't mean others know how to overclock. I'm particularly against advising new users to overclock, due to their relatively less understanding and experience with the computer architecture. What if they fried their chip? What if they fried their motherboard? Are you going to be responsible for advising them into overclocking their computers in the first place?

Secondly, the motor analogy is just poor. Overclocking, and especially overvolting significantly raise the power consumption and heat dissipation. Using your favorite Q6600 G0 as an example: TDP for Q6600 at stock is 95W, while TDP for Q6600 at 3.6Ghz, 1.4V is 179W. Is that "little change in gas mileage"?

The better analogy would be purchasing a 200hp motor, and tweak it to 250hp output, at the expense of significantly increased fuel consumption, as well as engine wear.

Again, my point is that while you may have substantial knowledge in overclocking (or as you claimed), doesn't mean others can do the same.

That's actually incorrect. I've successfully overclocked a Q6600 to 3.87Ghz. Don't get me wrong here, I have nothing against overclocking. But advising overclocking to new users is just dumb and irresponsible.

Anybody on the shared 130 TDP on Bloomfield? Is that FUD? Or does it play somewhat into what DS is spraying?

^heh.. like I've mention before, Dragonsprayer is like a talking advertisement rather then give helpful advice. I wouldn't take anything DS has to say seriously, but then again, I kinda do when someone is asking help rather then display what he does for a.. perhap part time living?

True that. This is also another point I would like to stress. 3.6Ghz IS NOT THE SWEET SPOT. From my experience (which may vary, but the general trend should be similar), 3.2Ghz requires a 0.05V bump to 1.3V, while 3.6Ghz requires an additional 0.15V bump to 1.4V.

How is it a sweet spot if only 0.05V required to obtain a 800Mhz overclock, and additional 0.15V to achieve the last 400Mhz?

I believe the sweet spot, where lowest overvolting results in maximum overclocking, is around 3.2Ghz.