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
segagenesis - Thursday, April 14, 2005 - link
#44 - I dont think its *that* silly to say such a thing. DDR2 and PCI-E are still new technologies and apart from newer mainboards coming with onboard PCI-E gigabit lan, there hasnt been anything worthy of note for the mainstream user. Getting off the PCI bus is good but it takes time for us to migrate to it. Let alone were talking about technology thats barely penetrating the market thats already saturated with people who are perfectly happy with thier current systems. Remember how long it took for us to get off ISA completely.There are alot of 2-3ghz PCI systems out there and to Average Joe User (tm) you can spin PCI-E as much as you want but unless they are in the market for a new computer they really dont give a damn. Same thing for Athlon 64 or Pentium D. How do you convince someone who uses AOL they need THAT much power?
Food for tought.
Anand Lal Shimpi - Thursday, April 14, 2005 - link
MotleyI agree that there are huge benefits to PCI Express, but for pretty much the entire life span of the 925X/915 platforms none of PCI Express' potential was even remotely tapped into. So here we are today, where PCI Express devices are finally starting to appear and we are given a brand new chipset, one that supports dual core.
I didn't mean to come off as saying that PCI Express and DDR2 are bad technologies, but the 925X/915 platform as a whole was not aided by their inclusion during its life span. The 955/945 chipsets will succeed in those areas where the 925X/915 did not, although it is worth pointing out that while Intel remains on a 800MHz FSB - DDR2 continues to do nothing for performance, even on 955X.
Take care,
Anand
Motley - Thursday, April 14, 2005 - link
"Honestly, Intel processors and even the platform haven’t been interesting since the introduction of Prescott. They have been too hot and poor performers, not to mention that the latest Intel platforms forced a transition to technologies that basically offered no performance benefits (DDR2, PCI Express)".I find it absolutely disturbing that even anandtech would say something as silly as this. Sure, if all you care about is graphics performance, PCI-E isn't that big of a deal. But drop in a Gigabit ethernet card, SCSI controller running a fast/wide raid, or *gasp* iSCSI. You'll see the difference immediately. To say there is no performance benefit is just totally missing the point that PCI-E is an improvement over PCI. Get your head out of games for a minute and you'll see.
anandtechrocks - Thursday, April 14, 2005 - link
For someone who is desperately trying to prove that he has the biggest “IT knowledge Penis” you sure do come across as a profane school child overclockingoodness.segagenesis - Thursday, April 14, 2005 - link
Now now everone, dont stoop down to his level.Granted when you talk about how everyone who reads this site are a "bunch of idiots" and then talk about how vast your knowledge of IT is all the while not being able to back it up, this just shows immaturity and makes it hard to believe what you claim is even true. And yeah, when you have to stoop to correcting minor spelling errors (when you yourself were in the wrong) to prove a point it means you havent got a leg to stand on.
Not to mention having a fanboyism on something is actually a BAD thing to have in the IT industry. If you are so well intwined on a certain piece of hardware (say, all Intel and f*** everything else) then its a dangerous situation where you wont be trying alternatives. Possibly cheaper or better alternatives at that. Say your in a company and you have a specific solid mindset on something, Joe Blow 2.0 comes in as a new hire. Joe Blow 2.0 pitches cheaper, faster, better solution but you diss it because there cant be possibly anything better than what you have. Joe Blow 2.0 wins a contract and you look stupid. Don't try this at work kids.
The only defense I imagine he could possibly conjure up right now is currently in the market there is the "Nobody got fired for buying Intel" mentality where companies and such are wary of trying non-Intel products mainly because... Dell and other major manufacturers wont offer it in any quantity. All the systems here are Dell and its sometimes its a blessing and a curse. Of course, even the "Nobody got fired for buying IBM" mentality slowly faded so it depends on what the market wants.
Just dont criticize all pc enthusiasts because they want something other than the norm, thanks Questar.
Rapsven - Thursday, April 14, 2005 - link
"In 2005 I will purchase 11,500 desktop/notebook systems, and 900-975 servers."So basically, if we were to take that piece of an ego boost for a fact, you're probably an arrogant executive who thinks he knows everything and nobody can prove him wrong.
In reality, you're probably just a 16 year old "know-it-all" who has to be a grammar Nazi to prove your stupid little points that don't really mean anything in the first place.
For the moment, NVIDIA and Intel have a cross-licensing contract, so it's basically eye for an eye. Intel gets SLI, NVIDIA gets to make chipsets for Intel based systems. Since none of us actually know the exact specifications of the contract, I guess we can't make any comments on that, can we? But if NVIDIA eats up Intel's marketshare in chipsets, it's definitely a problem for Intel.
Go away, your comments are worthless.
overclockingoodness - Thursday, April 14, 2005 - link
#38: I definitely agree. LOLAfter reading Anand's comments, it looks like Questar doesn't know anything he claims he does. Apparenlty, you've got to learn about business as well.
Take care Dumbass and have fun wasting your company's funds on Intel CPUs.
overclockingoodness - Thursday, April 14, 2005 - link
I hope Anand's comments shut you up Questar.mlittl3 - Thursday, April 14, 2005 - link
#35,Man, I feel sorry for the company that is going to get all the crap that Questar is going to buy.
Maybe he works for Transmeta. :)
overclockingoodness - Thursday, April 14, 2005 - link
#34: By worthless I do mean that they are not worthy enough to consider for high-end performance. I was refeering to the enthusiast community than anything else.#35: It's quite unfortunate that you work in the industry. So, let me guess - all of these servers/laptops/desktops will have "Intel Inside". I guess it's quite stupid of the management of your company to give you such a huge responsibility, since you obviously don't know anything about it.
And if you own your own company then I can only wonder when your company will go down.