The Ultimate Hardware Guide [Last Update: 4-14-06]

[levicki]

My rig is an absolute bitch to OC. I can't get above 2.34Ghz. I've narrowed the problem down to 2 things:

1. my PSU is underpowered. it's a 485w enermax and i'm running 2x GTX's in SLI. I think by OC'ing i'm asking too much of it in terms of voltages.

If that ener is not single-rail one, you can try with one that is. I have heard that both SLI and Crossfire have problems with dual-rail PSUs. Makes one wonder why did they made ATX 2.0 spec at all if it doesn't do any good?

 

[Atolsammeek]

Big deal Fact remains. Anything you buy today will be out dated In a month or so. I m waiting for the Newer Amd chip. See you dont understand somepeople are on the road vers sitting on there rear at home.

 

[MadModMike]

Hmm...Okay?

~~Mad Mod Mike, pimpin' the world 1 rig at a time

 

[MadModMike]

Update. Review 1st post for changes.

~~Mad Mod Mike, pimpin' the world 1 rig at a time

 

[MadModMike]

Update 2. Review 1st post for changes.

~~Mad Mod Mike, pimpin' the world 1 rig at a time

 

[mpjesse]

Mike and I pretty much determined it's the PSU (it's by no means bad, it just doesn't have enough wattage). I'm ordering a new one tonight actually.

-mpjesse

 

[mpjesse]

FYI... i just bought this whore of a PSU.

http://www.newegg.com/Product/Product.asp?Item=N82E16817194004

/me luvs enermax. ;-)

I'll let you know how it turns out... if you'd like.

-mpjesse

 

[MadModMike]

Yea, for sure tell me how it went. BTW, that is a slick PSU.

~~Mad Mod Mike, pimpin' the world 1 rig at a time

 

[MadModMike]

Update. Review 1st post for changes.

~~Mad Mod Mike, pimpin' the world 1 rig at a time

 

[MadModMike]

Update. Check 1st post for changes.

~~Mad Mod Mike, pimpin' the world 1 rig at a time

 

[Hameedo]

Mike , can you tell me about the perofrmance gain of supporting 64-bit on any processor ...I mean how does this benefits normal 32-bit applications ..why they are called 32-bit app ?

also when I say that a certain bus is 128-bit (eg, GPU cores which is 256-bit ) what does it mean ? ..does it mean that it can recieve larger amount of data ? yes ? but how ?.

sorry if I asked too questions , I hope you answer them simply ...

 

[MadModMike]

Mike , can you tell me about the perofrmance gain of supporting 64-bit on any processor ...I mean how does this benefits normal 32-bit applications ..why they are called 32-bit app ?

also when I say that a certain bus is 128-bit (eg, GPU cores which is 256-bit ) what does it mean ? ..does it mean that it can recieve larger amount of data ? yes ? but how ?.

sorry if I asked too questions , I hope you answer them simply ...

1) Using a 64-bit OS (obviously with a 64-bit CPU) I have noticed and documented personally perfomance gains of up to 12%with normal operation and running 32-bit applications. The entire gain of 64-bit over 32-bit can be up to 75%. Also, 32-bit applications (32-bit apps = applications) benefit because running 64-bit increases Memory Bandwidth and CPU performance which in turn increases the performance of 32-bit applications such as Games and 3D Rendering programs.

2) When it says 128-Bit Memory Bus, it means that's the length of the Bus, it's 128 bits wide. This means it increases performance beacause it can send more information per clock. An example is that a 256-Bit Memory Bus will, theoretically, offer double the Memory Bandwidth of a 128-bit bus. The larget bus's (wider) allows more information to flow in and out at a time, but this also creates latencies because of how large the bus is and how fast the clock speed of the bus is relative to the size.

If you need more simpler responses or want to know anything else, just ask. I don't have a problem helping.

~~Mad Mod Mike, pimpin' the world 1 rig at a time

 

[Hameedo]

so when I say that a Memory chip is clocked at 1000MHZ and 128-bit bus and another memory chip with same frequency but has 256-bit bus ..
which is better ?

the number of bits is related to the bandwith ..right ?

 

[Hameedo]

another thing about 32-bit applications ...we all know that 1 Byte = 8 bits ...so we have 32/8 = 4 Bytes ..
so 32-bit applications means that they have 4 bytes per cycle 8O or what :?: these things are really confusing me !

 

[MadModMike]

another thing about 32-bit applications ...we all know that 1 Byte = 8 bits ...so we have 32/8 = 4 Bytes ..
so 32-bit applications means that they have 4 bytes per cycle 8O or what :?: these things are really confusing me !

I will respond to both of your posts with this post:

1st Post: The 256-bit Memory bus @ 1000MHz will be double the memory bandwidth of the 128-bit, theoretically that means it should be twice as fast, but just as more GHz doesn't meacaln a faster CPU, a wider bus doesn't always mean more performance, but it does have greater memory bandwidth. Yes, bits is related to bandwidth, it's Bits x Frequency / 8 = Memory Bandwidth.

2nd Post: The 32-bit means that is how large the data size is it's transmitting. So yes, it is transmitting 4-Byte size information vs. 8-Byte size information. Theoretically, 64-bit creates a larger latency than 32-bit, as is the case with almost everything. It goes alot more complex and in detail, but that is the jist of 32-bit and 64-bit.

~~Mad Mod Mike, pimpin' the world 1 rig at a time

 

[Hameedo]

so normal processors have 32-bit FSB ..and 64-bit processors have 64-bit FSB ? ...
if that what it is ..in INTEL's processors why having the 64-bit as extensions only (EM64T) ...what is the difference between 64-bit controllers in INTEL and AMD .

 

[MadModMike]

so normal processors have 32-bit FSB ..and 64-bit processors have 64-bit FSB ? ...
if that what it is ..in INTEL's processors why having the 64-bit as extensions only (EM64T) ...what is the difference between 64-bit controllers in INTEL and AMD .

The 64-bit processors refers to the size of the registers, a register is just a portion inside the computer that it uses for storing temporary information and calculating, for example:

The # 10 needs to be added to the # 15.

1) CPU places the #10 into a register
2) CPU places the #15 into a register
3) CPU calculates it and places the #25 into a 3rd register
4) CPU deletes information from the 3 registers and outputs the #25 into the Video Card or wherever it needs to go

The Front Side Bus has nothing to do with a 64-bit or 32-bit CPU. The larger registers mean it can hold more information and process more at a time, and vice versa for smaller registers.

~~Mad Mod Mike, pimpin' the world 1 rig at a time

 

[Hameedo]

so the registers is not related to processor cache ? ..
I mean the cache could have 32-bit or 64-bit bus ?

 

[MadModMike]

so the registers is not related to processor cache ? ..
I mean the cache could have 32-bit or 64-bit bus ?

If you're referring to Level 1 and Level 2 cache, yes that is also unrelated to a 32-bit or 64-bit processor.

~~Mad Mod Mike, pimpin' the world 1 rig at a time

 

[Hameedo]

thanks Mike ..you've been a great help .

 

[ak47is1337]

The best valued overclocking chip from Intel may actually turn out to be the 805D rather than the 955EE. It seems that the only difference between the 805D and the rest of the 8xx series are that the 805D couldn't run stable using with a 800MHz FSB. All the other features are still in place including EM64T, 2x1MB cache, and EIST downclocking to 1.86GHz. The great thing is they may actually be excellent overclockers with the limiting factor being how close you can get the FSB to 800MHz. HKEPC got the base FSB up from 133MHz to 185MHz for a 3.7GHz processor. Not bad for a $150 processor. Of course, it hasn't been released yet, but I thought the schedule was for the end of February.

http://www.hkepc.com/hwdb/pd-805-1.htm

For MadModMike, I know others have already said it, but the guide is great. I do have a suggestion for the Intel dual core recommendation though. Personally, I find the 8xx series a better value right now because Intel recently dropped their prices to clear inventory. The best value is the 830D which retails for $277.90 putting it at the same price level as the 920D.

http://www.zipzoomfly.com/jsp/ProductDetail.jsp?ProductCode=80831&affiliate=shopping

I know you recommended the 9xx series because they run cooler, but the power and temperature difference isn't that large right now because EIST support in the 9xx series is disabled preventing downclocking when idle.

I don't know if you want to mention steppings since it may to too complicated for an entry guide, but the preferred stepping for the 8xx series is of course the latest which is B0.

You have more experience with AMD than I do, so I really can't recommend anything for that section. I hope the moderators make this a sticky.

Nah, 920's are excellent overclockers for not much more money. The fact they are on 65nm helps very, very much.

 

[chime]

Mr. MadModMike,

Can you shed some light on why people go into longer pipeline (ie Intel)? Being an analog guy, I am pretty much oblivious about digital stuff. I just remembered about 15 years ago of a discussion CISC vs. RISC and ,deep pipeline was a characteristic of RISC where as the software compiler must rearrange the data/command to maximize the pipeline? Is this true?

Thanks.

Chime.

PS: Fanboy flaming is fun to read :lol:

 

[dvdpiddy]

dude pm him if you want i dont think he reads this thread anymore

 

[corvetteguy]

dude pm him if you want i dont think he reads this thread anymore

dude, wasn't this the thread he wanted to make a sticky? Why wouldn't he read it anymore. :?

 

[dvdpiddy]

cause he probably for got about it