Dual Core Intel Platform Shootout - NVIDIA nForce4 vs. Intel 955X
by Anand Lal Shimpi on April 14, 2005 1:01 PM EST- Posted in
- CPUs
SATA Controller Performance
Both NVIDIA and Intel offer support for NCQ in their SATA controllers, and given our recently renewed interest in NCQ performance, we decided to find out if there were any performance differences between the two SATA controllers. However, as we've found in the past, coming up with tests that stress NCQ is quite difficult. Luckily, there is a tool that works perfectly for controlling the type of disk accesses that you want to test: iometer.
An Intel developed tool, iometer allows you to control the size, randomness and frequency, among other things, of disk accesses, and measure performance using data generated according to these specifications. Given that NCQ truly optimizes performance when disk accesses are random in nature, we decided to look at how performance varied according to what percentage of the disk accesses were random. At the same time, we wanted the tests to be modeled on a multitasking desktop system, so we did some investigation by setting up a computer and running through some of our multitasking scenarios on it.
What we found is that on modern day hard drives, the number of outstanding IOs (IO Queue Depth) is rarely above 10 on even a moderately taxed system. Only when you approach extremely heavy multitasking loads (heavier than anything that we've ever tested) do you break into queue depths beyond 32. So, we put together two scenarios, one with a queue depth of 8 and one with a queue depth of 32 - the latter being more of an extreme condition.
In each scenario, we sent the drives a series of 64KB requests, 75% of which were reads, 25% were writes; once again, derived from monitoring our own desktop usage patterns.
We then varied the randomness of disk accesses from 0% (e.g. 100% sequential) up to 100% (0% sequential reads/writes). In theory, the stronger NCQ controllers will show better performance as the percentage of random accesses increases. We reported both Average IOs per Second and average IO response time (how long accesses took to complete on average):
With a queue depth of 8, the two SATA controllers offer virtually identical performance.
Looking at latency, Intel actually offers a very slight performance advantage here - nothing huge, but it's definitely there.
The results get much more interesting as we increase the queue depth to 32:
Here, NVIDIA starts to pull away offering close to a 20% increase in average IOs per second as the access patterns get more random (e.g. as more applications running at the same time start loading down the hard disk).
What's truly impressive, however, is the reduction in average response time - up to a 90ms decrease in response time, thanks to NVIDIA's superior NCQ implementation.
But stepping back into reality, how big of a difference NVIDIA's NCQ implementation makes depends greatly on your usage patterns. Heavy multitaskers that are very IO bound will notice a performance difference, while more casual multitaskers would be hard pressed to find any difference. For example, Intel was actually faster than NVIDIA in our gaming multitasking scenarios from our dual core investigation.
Facebook
Twitter
RSS
96 Comments
View All Comments
KeithDust2000 - Thursday, April 14, 2005 - link
JoKeRr, "However, with such a great product, why couldn't AMD strike a deal with major OEMs like Dell? I know they're doing great on the server side with opteron, but why not desktop?"You must be "joking". Dell may be the the biggest PC OEM in the world, but not by much, and No.2 HP, No.3
Lenovo/IBM, No.4 Acer and No.5 Fujitsu-Siemens all carry AMD desktop CPUs, from Sempron to Athlon 64.
chennhui - Thursday, April 14, 2005 - link
Quote:-Why not impliment something like Hyperthreading when it's proven to work so well?
" Fred's response to this question was thankfully straightforward; he isn't a fan of Intel's Hyper Threading in the sense that the entire pipeline is shared between multiple threads. In Fred's words, "it's a misuse of resources." "
http://www.anandtech.com/cpuchipsets/showdoc.aspx?...
KeithDust2000 - Thursday, April 14, 2005 - link
Hi Anand, wrt. your comment"A bit of that changed when Intel brought forth their dual core plans - assuming that they can actually guarantee availability, Intel is planning to ship more desktop dual core processors, at lower prices, than AMD this year."
I think it´s unlikely that the 1.8Ghz Athlon 64 X2 will cost more than $240. After all, a single core Athlon CPU @ 1.8Ghz can be had for ~$125. Not mentioning that it will likely show higher overall performance (note that Pentium D has no HT), will fit into existing platforms, have about half the power consumption, and probably better 64bit performance and power management. We´ll see, I guess...
JoKeRr - Thursday, April 14, 2005 - link
OK, I was going to flame Questar with an old Chinese saying:It's always the idiot who got the loudest mouth and who thinks he/she knows everything, and it's always the smarties who only speak the truth and state the facts when necessary.
Anyway, I take that back b/c I think Questar has changed.
-------
On the flip side, this is how I view the whole processor debate:
Intel:
great marketing power and OEM support, great fabs and lots of engineering power. P4 prescott is hotter than it should be, but Intel has made a lot of progress from C0 stepping to the newest N0 stepping. Plus you've got to give Intel some credit for being the first that brings Dualcore to desktop (I know AMD's the first in server) and willing to widely spread the eventual benefit of dualcore to everyone today at a very fair price even though the cost of manufacturing is definitely higher. Sure it's essentially 2 prescott 1mb core glued together but it works. It takes guts for a company to widespread something so bleeding edge and so much at the same time (DDR2, PCI-E, HD audio, etc). And if we look ahead, Intel is making dualcore 65nm's thermal evenlope same as Northwood at 89W (don't count on me though), I'm impressed b/c Intel is actively addressing the problems. And who's not impressed with the introduction of Centrino?
On the other hand, there are things I don't like about Intel either: Such as the frequent change of socket (was 775 really neccessary before dualcore?? what about 423?), not supplying nearly as much 875P chipset as it should b/c 925XE/925/915 is not selling well, and now a whole new platform just to add dualcore support, just to name a few. As for performance: Other than gaming, Intel's P4 is only behind in 3 or 4 benchmarks when I last counted.
AMD: They do have a wondering processor with Athlon64 right now, cool and fast, especially in gaming. They are also making good progress with dualcore, and mobile platform (turion64) and they have great chipset from NVDIA and VIA. And props to AMD who's the first to introduce 64bit support, well done. However, with such a great product, why couldn't AMD strike a deal with major OEMs like Dell? I know they're doing great on the server side with opteron, but why not desktop? Lack of marketing in my own opinion hurts a lot for companies like AMD. And I wish AMD could have addressed the issue of multitasking better before the coming of dualcore: Why not impliment something like Hyperthreading when it's proven to work so well?
---------
One last question for Anand b4 I shut up:
DDR could run at 2-2-2-5 but fetches 2bit per cycle. DDR2 runs at 4-4-4-10, even though twice the latency but fetches 4bit per cycle, so essentially DDR2 at 4-4-4-10 is about the same (in terms of bandwidth and maybe even latency??) as DDR at 2-2-2-5?
Anand Lal Shimpi - Thursday, April 14, 2005 - link
minoI'll see if I can get Derek to do an article on audio quality of the latest solutions, but a lot of that will vary from one motherboard to the text. If an article does end up shaping out, I'll post something about it.
The 2 single core CPUs vs. 1 dual core CPU comparison is an interesting one that I'd like to make and I'll do my best to fit some of those numbers in there, but I think there are other, more useful (from a purchasing standpoint) comparisons out there that you will see in the article.
As far as my reasons for not doing an AMD NF4 vs. Intel NF4 comparison, it has nothing to do with pleasing any manufacturer - as I've said many times before, I don't care about pleasing any manufacturers, I'm here to deliver what you all want. It doesn't matter that Intel supplied the CPU, they just send us the hardware and we do whatever we want with it. I was originally going to do an AMD vs. Intel comparison in that article, but a handful of readers responded that it wasn't necessary so I left it out - I agreed with them as I thought it would be redundant and after all, if you're looking at a comparison of Intel chipsets you've already decided that you want an Intel processor (if not, consult our CPU reviews first to figure out what CPU to buy, then read the chipset reviews to figure out what chipset, then what motherboard, etc...).
Take care,
Anand
mino - Thursday, April 14, 2005 - link
Just one correction that's worth it:" Albeit not beeing unjustified. At least insom cases like HD audio. "
should sound more like this
" Albeit not beeing justified. At least in some cases like HD audio. "
You know, in my native language there could be even 3 to 4 negations in one simple sentence:
Nikdy som nepovedal ze nie si blbec. goes word for word:
I never didn't said that you aren't silly.
Funny those "dirty" translations are. In manuals topic it's sometimes an issue if instead of EN you've got just your native one.;(
Soo, thats all for some uninteresting dully comments. At least have got a feeling I wasn't alone...
mino - Thursday, April 14, 2005 - link
bleh; just got pissed of by my own spelling/hypnetation/thesaurus/etc .. ;)mino - Thursday, April 14, 2005 - link
Rand: Yeah ! so am I.Actually this is the first time I noticed they have changed it ;-) Funny is I started to watch IT closely just 2yrs bacwards...
Questar:
from the evolution of your posts through out this discusion I see you HAVE changed the opinion. It's quite refreshing to see some valid arguments in your last post. That's it you should have started with.
In case you are who you claim to, this flame was surely worth the paper (literaly meant :).
Anand:
Please could you take a look at the audio quiality of the new SB from nVidia(;-) ? You know, nF4's AC'97 is nothing to sing about...
From the other keg - hope to see 2x248(848) versus single 1x275(875) in the SAME board compared. this would be waaaays more usefull spent time than any other comparison possible. PLEASE take note here so those of us who have current image of situation in the performance arena doesn't have to make indirect guesses(wonder if that word spells such way ;D).
To all who would like to se nF4 AMD vs. Intel SLI roundup: Please take in mind that such an embarassment would not please Intel(AND the provider of tested HW) very much. It would also make no other sense than to sink the P4 platform even more into mud. Albeit not beeing unjustified. At least insom cases like HD audio.
Rand - Thursday, April 14, 2005 - link
Kristopher-I still spell it as nVidia all the time, purely out of habit. Even after all these years I haven't quite gotten it stuck into my head that NVIDIA is now the appropriate spelling.
Rand - Thursday, April 14, 2005 - link