AMD vs Intel

I am looking at getting a new laptop, but can't decide between AMD or Intel, anyone who can help, I need it to remotely log on to company server, as to do maintenance on the Database - will have to work from some very remote locations including northern Mozambique - any help will be greatly appreciated

Depends on how much money you want to spend. As (very) rough guideline, the AMD and Intel CPUs correspond roundabout so:

AMD Sempron <--> Intel Celeron-M
AMD Turion64/Mobile Athlon64 <--> Intel Pentium-M/Core Solo
AMD Turion64_X2 <--> Intel Core Duo
????? <--> Intel Core2 Duo

Core and Core2 are not the same and done mean one core or two. Core the architecture, Core2 is it's newer version. You get both Core and Core2 in sinle (Core Solo/Core2 Solo) and dual (Core Duo/Core2 Duo) cores (yes, intel is again trying to confuse the living daylights out of unsuspecting users).

If you're strickly speaking CPU performance, take the AMD Sempron/Turion64/Turion64_X2 over their Intel counterpart. They are faster, especially the Sempron, if you're in that price range. They're really quick for the money.

AMD don't yet have anything that rivals the Core2 duo. If you can afford that, go for it - it will last you a while.

I have a Turion64 notebook. If I have to buy a new notebook now, I'll probably buy an Intel, for no other reason than the sheer variety of notebooks available. It's much easier to find an Intel based on that meets all my criteria than an AMD one. Right now the only AMD one I know of that has all the goodies is the Ferrari 5000 (which last time I checked isn't available here yet). I'm sure there are more, but they're all in that stratospheric price range where I'd much rather buy myself a Macbook Pro.

It's a shame that manufacturers put so little faith in AMD. My notebook is a 1.8GHz single core. In CPU intensive stuff - compressing large backup sets, compiling software - it comfortably outperforms our office server, which is a dual 3ghz Xeon (1mb cache per CPU).
 
It's a shame that manufacturers put so little faith in AMD. My notebook is a 1.8GHz single core. In CPU intensive stuff - compressing large backup sets, compiling software - it comfortably outperforms our office server, which is a dual 3ghz Xeon (1mb cache per CPU).

Come now. That sounds a little bit harsh? How can a 1.8Ghz laptop outperform a XEON Server?:confused:
 
Come now. That sounds a little bit harsh? How can a 1.8Ghz laptop outperform a XEON Server?:confused:

Ask Intel why a 2GHz Core 2 Duo can outperform a much faster clocked Pentium D. Or why Athlon 64 processors in the 2+ GHz range can thrash Pentium 4 processors well into the 3GHz range.

It's all about what you during in the clock cycles, not how many clock cycles you can squeeze in.
 
Come now. That sounds a little bit harsh? How can a 1.8Ghz laptop outperform a XEON Server?:confused:

The Pentium 4 and Pentium 4 based Xeons are way overrated. They're good with stuff like encoding video, but that's about all. Intel left out so many bits and pieces from that architecture in order to deliver on the GHz promise, that under certain circumstances, they compare poorly even to the Pentium 3.

For example. We have a client who runs a bunch of thin clients, all logging into a Windows terminal server. Anything 25-30 users at any given time, except luch time, when it drops to about 15. Server was a dual Xeon - 2.4GHz if I remember correctly, 2GB DDR ram, two 10000rpm SCSI discs in a RAID-1. Intel server board, chassis, server NICs, etc. Good hardware all round. They clients ran mostly MS Office, Pastel, MS Great Plains and MSN/IE and the usual distractions.

This particular server ran at 100% CPU usage, most of the time. The box was replaced with a similarly specced one, but different model mobard, differend RAID card, etc. Same load.

So they got hold of two secondhand Dell Poweredge boxes, identical ones. Dual Pentium 3 1.1GHz (Tualatin core), 1GB PC133 ram, two 10000rpm SCSI discs in RAID-1, server NICs, etc. So aprat from CPU, motherboard and memory, the machines are similar to the Xeon box. We setup both boxes identically, split the users half half and shared the load over the two boxes. Right, avarage CPU load was about 20-40%. The one of the two Dells packed up, and for the time being, we moved the other half of the users to the first Dell.

Now the same group of users that flatlined a dual Xeon are now sharing a Dual Pentium 3 (and not even a P3 class Xeon). Guess what, the avarage load hardly touches 80%. Two years later, they're still on the one Dell.

With regards to the Turion64 compared to the dual 3GHz Xeon - it's more or less like this. Both run 32bit linux. Both have 2GB memory. Both run a variety of database servers, including MySQL, postgres, and Cache. Both run apache, ftp servers, samba, and a host of other small business type applications. The notebook is my testing ground, and the comparisons was made in the wee hours of the morning, when the server is standing idle.

I never thought of comparing a notebook (any notebook) to the server, but we got a 1.6GHZ Sempron (64bit) destkop box that had to be set up for multiple-domain mail hosting. While working on this I noticed that it does certain things quicker than the XEON server. So that spiked my interest.

First, I compared kernel compile times (identical config, same version, both with -j4 - which should benifit the dual CPU box). The dual Xeon came out at 4 minutes, the Turion64 at 3. I don't remember the exact times but it was more or less that.

Then I took my mail directory (about 6 GB), copied it to the server, and compressed the whole bunch on both sides. They finished in roughly the same time, but remember the server has 10krpm SCSI discs, the notebook has a single 5400rpm IDE. If the discs were the same you can guess which one would have been faster.

Lastly, a programmer buddy wrote a little program that generates 4096bit RSA certificates (which is quite intensive) to compare the difference in 32 and 64bit processing on their new 64bit Xeon servers. As the developer said, "not the most academic and accurate test, but it gives you an idea of a real-world application." He doesn't have access to an AMD64 box so he couldn't have optimised it for anything but the Intel. The program generates 10 certificates, and gives you the total time. I ran each test 10 times and worked out an average. Here's what I found:

Dual 2.8GHz Xeon, 1mb cache, 32bit: 137 seconds
Dual 3.0GHz Xeon, 1mb cache, 32bit: 122 seconds (completely different box)
Pentium4 64bit HT 3GHz, 2mb cache, 32bit: 138 seconds
Turion64 1.8Ghz, 1mb cache, 32bit: 73 seconds
Sempron64 2600+ 1.6ghz, 256k cache, 32bit: 82 seconds

Now, throw some 64bit results into the mix:

Pentium4 64bit HT 3GHz, 2mb cache, 32bit: 51 seconds
Turion64 1.8Ghz, 1mb cache, 32bit: 46 seconds
Athlon64 2.2GHz 512k cache, 64bit: 29 seconds (I don't have 32bit libs installed on this box)

The Pentium 4/Xeon (netburst) vs AMD64 comparison kinda reminds me of that old Chomp ad with the two hippos. One bites off little bits at a time (and chews them quickly), the other swollows the whole damn thing in one byte.

Netburst has long been overtaken by the Pentium-M architechture. Spending money on any Netburst based hardware now is foolish - you can do a lot better for the money, even if you stick with Intel.
 
Top
Sign up to the MyBroadband newsletter
X