Intel P4 vs AMD

Actually the 2.8 Ghz Northwood can easily hit 3.2 Ghz I know I've done it and I'm about to do it again. I just ordered a refurbished Asus board for $61 that has dual channel memory capability and is an overclocking demon. I have everything else so all that I need is a case for $50. Total cost $110 for a 3.2 Ghz or faster Northwood (my old board didn't have dual channel).

 

Sophocles, if you're using the memory you had, it's better than what I have. You've seen the specs. I'd have to buy more parts as well. You have the power supply and case. It would almost be like building a system and scavanging drives, if I want to do it right. About the only things I'd be able to use would be the CPU, drives, cables and some of the mounting hardware.

I'd rather spring for the P4 3.4E and enjoy some upgrades. They're selling for around $280. My board supports much of the new technology, so no problems with it except everything is fixed. With the Extreme P4 I'd have HyperThreading as well as the dual channel I already have. With performance like that, I might even spring for the RAID 0 setup. LOL I've already got the storage. With that, I might last to nearly '08 before having to buy into the newer tech. Since I'm no big time gamer, I'm not worried about the graphics cards. A good AGP card does okay for me. In fact, I have no complaints with the old ATI Radeon card I have now. The P4 3.4E shouldn't be overclocked further, at least according to the Intel data, so a custom board wouldn't be of much help if I go the CPU only route with the 3.4E.

Feel free to correct me if you see a flaw in my reasoning or if I'm going astray somewhere. You're more familiar with systems than I am.

 

brobear

Your old memory will do just fine and so will everything else, and remember an overclocked chip is faster than a stock chip configuration.

 

On the memory I know, but 400MHz would be much better than the 200MHz I have. Also, the timing is better with the performance oriented RAM. What I have is just decent "off the shelf". Keeping down build costs was a factor in what I have. Off the shelf PCs are notorious for that. Luckily they didn't get too cheap on me; they installed 2 512MB cards instead of 4 256MB ones and I guess the timing could be worse than 3-3-3-8 at 200MHz (as we know, it could be better). I was under the assumption that for an OCed system to be faster, the faster memory had to be working along with the CPU and 400MHz was much better than 200MHz.

You didn't appear to want to push the the 2.8 Northwood over 3.2GHz. The P4 3.4E has a speed of 3.4 with the 800MHz FSB. In that situation the memory I have would work good. I was looking at the quality Corsair ($244 for 2x1GB matched and about $125 for 2x512MB CMX1024-3200C2, PC3200 400MHZ CL2 184-PIN DDR DIMM W/HEAT SPREADER ). I know, the 512s are faster, but if I ever wanted to try booting up Vista, I might want a lot more RAM.

Case, power supply, motherboard, RAM (optional), thermal compound, fans (maybe). That's to get to 3.2 GHz. To get to 3.4E is about $280 and it is an upgrade over the older Pentium which doesn't have the HyperThreading and the newer CPU doubles the cache to 1MB along with having the SSE3 Multimedia Instruction.

I know you like tweaking and having more control over your system settings. However, from the aspect of just having a good working PC with the most features with a CPU, which route would you suggest? As I said earlier, an AGP card for graphics is all I need, so that isn't a point of interest.

 

brobear 200Mhz memory is 400 Mhz or 800 Mhz memory. In an 800 Mhz front side bus the memory's native speed is 200 Mhz but it's quad pumped or 4 times 200. In my system its 5 times 200, and when I set my hypertransport speed to 4 times I raised my memory to 240 and now it's 4 times 240 or 960.

 

For general purpose speed raid 0 kicks ass.



These are just 2 seagate 80g sata drives with ncq.

XP installs completely in less than 10 minutes.
It has been a while since I used Raid 0. I forgot what I was missing.

 

I'm with brobear on getting a new processor. Intels prices are pretty good right now.

Being an AMD whore right now I would go for a 3500+ and motherboard for 250 dollars and keep the same memory case and powersupply. Just a thought, but more work.

 

Sophocles, possibly I got some numbers mixed up with max bandwidth and native speeds. I've still got a lot to learn about technical aspects of the components. I wasn't going back and reviewing the numbers. But as we both know, there's a difference between C2 and the C3 RAM I have and most people seriously considering OCing a system opt for the C2. 2-3-3-6 at 200MHz with C2 works out better than 3-3-3-8 with the C3. With the C3 I only get 2.5-3-3-7 at 166MHz.

 

Donald
If I kept the same case, I'd have to do cooling mods as well. I don't know if you've looked inside some of the Dells, but they have a proprietary set up with a cooling duct running over the processor. If the boards don't set up to put things in the same spot, the duct isn't doing the job. That means some additional cooling work and taking out the duct or at least modifying it. Going with the P4 3.4E is also the lazy way of doing the job. LOL

On your benchmark, what drives are you using and what size and RPM are they? Good time, that's twice as fast as my 7200 RPM WD on my PC.

 

Sophocles,
I noticed in your post to sammorris that you own an Asus P4P800SE MB. Now I know how you came up with all the excellent advice for mine. It still runs like a dream OC'd from 3.0 to 3.40. I just did a little math and discovered that if I build my new dual-core system today it would cost me $229.00 for the MB, $618.00 for the video cards, $497.00 for the 4400+ CPU and $312.00 for the 2 Raptor 10,000 rpm 74 GB SATA hard drives. This would come to $1656.00. Hopefully these prices will go down by mid year. For now I'm just going to shop sales and see what I can get put away til then. So far I've set pretty much in stone which MB, Video cards, CPU and Hard drives. Since I know nothing about Raid, SLI or SATA drives I know I'll be bugging this forum with all sorts of questions.

theonejrs

 

If you take note at 64026402's posted results, you'll see that even that RAID setup still doesn't saturate the ancient ATA100 spec. Yes, burst speeds are alot better on those setups, as they can be with alot of single drives. However, your best bet for drive performance is not from a RAID setup, but from drive speed, and onboard cache.

If someone can, set a test using a single drive, then compare that with a test of a RAID setup using the same two drives.

For example: Test one 120gig drive (7200rpm) with an 8mb cache. Then, pair it up with another identical 120gig drive (it's twin for the sake of the convo (in RAID))...run the tests several times (3x should be your mine), and average the results.

Then....after all's said and done, test it against a 7200rpm 16mb cache drive.

I'm willing to bet that you will see a better performance hit from a single 16mb cache drive, then the other two results.

I wish I had kept the MaxPC mag that had this very test in one of their issues, cuz the proof was in the pudding as they say.

You can even get some interesting drive results using Nero's DriveSpeed on a virtual drive. You'll see the burst speed, then the actual speed of it's read/write caps.

 

That's a skewed test. Why not finish it. The test should run with the 120GB 7200 RPM 8MB cache tested singly and in RAID 0 configuration. Then the 120GB 7200 RPM drive with the 16MB cache should be tested both singly and in RAID 0 configuration. That way you get a more complete picture, difference in results depending on cache, improvements over single drives in RAID for both cache levels, even the difference between 8 and 16MB cache drives in Raid 0 configuration. The magazine didn't quite finish what they started.

What they showed was how a better drive matches up to an inferior one with a RAID 0 configuration. What they should have shown was how two drives compare in both single and RAID 0 configuration. Makes quite a difference in how those tests are set up. You have to be careful when you look at experiments others set up. It can skew the info in ways that leads to false assumptions.

By the way, note the size of the ATA 100 drives are 80GB in a RAID 0 setup. That's why I asked the drive size and speed on Donald's system. The 80GB drives in ATA 100 RAID 0 are still one heck of a lot slower. I don't see the reasoning. All it shows is that the ATA 100 setup is much slower than Donald's. Little else can be told from the test. Interesting is that a SATA 150 set of 36GB drives are running at the same speed. Drive size, free space, fragmentation, there are all kinds of variables that can skew tests like this as well. Remember it's timing access to data and that depends on the drive size, speed, configuration, and the condition.

 

personally I think the performance of my RAID array is limited by it being split into diffferent partitions using Acronis, since I didn't exceed 80MB/s in benchmarks of sequential reading.

 

I wouldn't partition a drive I was going to use in a RAID 0 setup and since all that would be on them would be the OS and programs, they wouldn't have to be very large.

 

If I remember correctly, the RAID setup was using Raptor drives. Unfortunately, the Raptor will die a horrible death, seeing as WD isn't going to make any larger than 74gig drives. (last I heard)

I really need to save the info I find!

 

You can achieve a similar effect by combining two higher storage drives now technology has improved so much.

 

Morph416

It's unlikely that the Raptors will get much larger because it would require adding more platters but you can put three or more in a RAID 0. If you do a little research you'll discover that Seagate ncq drives appoach Raptor speed in most areas for a whole lot less money. Add two of them in a RAID and they're going to prove to be a better deal than two raptors because you get most of the speed and half the price and more space. For those individuals that are really looking to speed things up my advice would be to get more RAM and disable the paging file. Usually 2 gigs is enough to do that.

 

so do the western digitals i have,match the speed of the raptors....those raptors are beginning to be old school,sata 2 is where its at but the 10,000 rpm drives are still real exspensive right now....if sata 3 comes out then they all will drop,always happens.......had to give my 2 cents ,not worth much but...lmao...peace

http://img350.imageshack.us/img350/2983/capturewiz0034pt.jpg

theres my mobo if it worked...got it thx brobear

http://img350.imageshack.us/my.php?image=capturewiz0046qh.jpg

theres my pc

 

Two of them in a RAID should beat an individual raptor in speed.

You've almost got your imaging down.

 

http://img42.imageshack.us/my.php?image=capturewiz0055dl.jpg


this part is easy ...dont understand the url part soph,im getting there though...thx guys,this is pretty kool


" [url]http://img42.imageshack.us/my.php?image=capturewiz0055dl.jpg[/url] border=0>

dont get in between you told me dosnt work...lmao