Corsair XMS DDR434 Ram

http://www.bit-tech.net/review/156/

 

I just got my Corsair DDR (400Mhz though)
But it was only $79.99 LOCALY (Fry's electronics sale)
For XMS 3200 2 Cas. Platunum.
That was cheaper than pricewatched lowest price!
Running AWSOME!!!

I can only imagine 434 works even better!

 

My system is crying for some top notch RAM, I'm really going to have to buy some of this soon

Rob.

 

Can't say what a top notch review this was for me. I learned a lot about the RAM/CPU relationship that makes a lot of sense. Thanks for spelling all of that out in the review. I have the XMS 3000 memory and couldn't be happier. Now, to get my hands on a Nforce2 board and an unlocked AMD chip

 

Try a different mobo before judging the ram pal.

I have an IT7 Max2 Ver 2 runnning a P4 at 3.2GHz with the FSB at 177MHz and the ram at 2-5-2-2 at over 220MHz and it is stable with the RAM at 2.8v.
So it seems that there is some variability in the quality of the ram. Oh and for your info I have 2 x 256MB sticks too. Seems that my system puts a whole lot of doubt on the worth of your reveiw.
Maybe you could have commented on the quality of your mobo before making sweeping statements about a product.
Also seems that my RAM is fater than PC 4200 when benched with Sandra.........

Do your homework.

Seems like the world is full of incompetent reveiwers these days.
Regards
ATH

 

You can be arrogant, you can critisise constructively; what you can't be is insulting.

Goodbye.

 

Re: Try a different mobo before judging the ram pal.

Thanks for your comments. It's always a pleasure to receive feedback. I am glad your are pleased with the memory.

I am puzzled as to your point as there doesn't seem to be one. Nor do you appear to have actually READ or understood the review.

The review did not attempt to achieve maximum memory bandwidth scores but try and assert the usefulness and worth of the product. If I wanted to impress with benchmarks then I would have tweaked the system to hell along with modifying the motherboard. Instead, I prefer to use untweaked systems which produce results realistically achievable by most people and as such representative of real world expectations. All products have some degree of variability in absolute performance.

Oh and for your info I tested a single stick of 512Mb not 2 x 256Mb.

Do your homework.

 

Hello macroman,

I'm afraid I also feel that your review left a bit to be desired. Though, if this is your first review you did a great job.

But running one benchmark really doesn't tell one anything. It only tells you how well your system runs that benchmark. And the results are often very misleading.

Generally testing with real world programs returns more useful information and give you a much better idea of what performance you will actually see.

A more useful benchmark is weighted average frames per second in popular games. Just make sure you also include minimum fps, and maximum fps.

Other applications besides games are useful for those who will do more than just play with their computer.

Maya Personal Learning Addition is available for free download, and would give you an example of how a system runs a professional OpenGL app. Though, as with all programs, different applications will react very differently to changes made in a system. That one of the reason many reviews use so many applications this is an attempt to give the reader an idea of what areas a product excels in.

Unfortunately one page reviews are generally not worth much of a read. The whole review can easily be summarized in one sentence.
RAM installed easily at 200Mhz, a bit little tricky towards 217Mhz, as Mhz was increase it returned somewhat better marks in Sandra benchmark.

Sorry if this is a bit harsh,
Again though, if this was your first review you did a really great job, my comments written with an experience reviewer in mind.

I'm especially sorry because I expect to be making a new PC within 30 days and I am trying to figure out if the a 2x512MB is worth the extra $200 over normal 2700 2x512MB.

StB

 

Ok it seems that the point of the review is being missed. This is PC3500 ram, of course it will be faster at that speed than say PC3200. The review was not about how much bandwidth could be squeezed out of the ram but how beneficial paying for a few extra MHz would be. The result, not much.

The fact I could have produced higher benchmarks by fine tuning the system is irrelevant. You will note the CPU used was an XP2400 clocked down to XP1600 levels! Also the review also says to disregard the scores as a measurement. They are meant for comparative purposes only and each set to be taken in isolation.

Quite frankly I am fed up seeing reviews that push hardware to the very limit and conclude that one product is better than another simply because it can run a few MHz faster. Manufacturing tolerences dictate that products, in this case memory, coming off the same production batch will have at least that amount of variation across samples. This I consider to be misleading to the readers who will spend their money expecting to achieve the same results only to be disappointed when they purchase a sample from the lower end of the tolerence scale.

As for using only one benchmarking tool, well if memory bandwidth improves then so will any application that is memory bandwidth reliant. I don't need to run the apps to prove that. Increasing the fps of a 3d game by 2 fps is hardly going to set the World on fire!

I try to cut through the marketing hype and techno BS and and find out if a product offers any real benefit. Corsair ram is an excellent product and one which I personally use and recommend but in this case the benefits from spending the extra cash are in real world terms non-existent.

As a side point, I left the reader to do their own analysis of the Sandra results and this was obviously a mistake , (inexperience showing here). The idea being you could draw your own conclusion as to the suitability of the ram in your own system based on the % diff between the various fsb/ram clock speeds. Maybe next time I should do the math for everyone to ensure the point is not lost.

If you are planning on running your memory at speeds above PC2700 then get the faster memory. At least you know it will do it.

PS Thanks for your comments.

 

Sorry didn't mean to be insulting.
I am a little blunt at times, one of my qualities. BTW thanks for the ban!
Maybe a warning would have been better?
Anyway it doesnt matter to me if you stick by your ban and ban me again, I only joined to comment on the article, but enough of that.

Well OK maybe the point of the reveiw was to show what is happening on the bandwidth front if the RAM is synchronous or unsynchronous with the fsb. For this the reveiw was well written and I enjoyed it.

I guess for a P4 system this is less important as the FSB is much faster so the bottleneck is always going to be the RAM. Maybe you could have pointed this out and clearly stated that the statements you made did not apply to a P4 sys, only to that of an AMD.

I guess it must bug me if ppl make comments about "PC hardware" and clearly mean either Intel or AMD products and fail to state so.

The RAM if run at default timings will OC to around 440 MHz. Corsair themselves state that the RAM is designed to run up to 2.8v and is still under warranty if this is the case. I have seen ppl get the RAM even higher. I still stick by the fact that the RAM is not a bad OCer as mine will not run at 217 default cas unless the Vdimm is set at 2.8v.

"This is PC3500 ram, of course it will be faster at that speed than say PC3200. The review was not about how much bandwidth could be squeezed out of the ram but how beneficial paying for a few extra MHz would be. The result, not much."

I gotta disagree in my case however if you aint got a system to run it then yes, it is a waste of money.


Anyhow thanks for the reveiw it was interesting and you can write much more competently than most reveiwers out there. Just felt that a couple of points were omitted. Plz accept an apology.
Regards
ATH

 

No worries, I appreciate people taking the time and effort to read and comment on my articles.

Since ram designed to run at higer speeds will obviously shift data quicker than lower speed memory, I chose to look at how different system fsbs would benefit from this particular increase in speed rather than how how rapidly I could shift data around.

As you say, this memory gives you significant improvement on your system. different people will get different results which is why I do not place much empasis on absolute performance with my test rigs. As to whether this increase in speed makes any practical difference to your computers performance, only you can say. I am yet to be convinced of any tangible benefit from running memory 10, 20, 30MHz higher for the computing majority. The laws of diminishing returns comes into play and I am more concerned with the overall computing expereience rather than isolated aspects of hardware performance. Maybe that's a trait of old age

Unfortunately I did not have a P4 system with which to compare, hence the lack of comparison although this should be soon remedied.

Thanks for your reply and I hope you continue to read our reviews and comment on them.

BTW It was not me who banned you, but I have unbanned you.

 

2 Important points here regarding the RAM itself and of the IT7MAX2. Firstly in order to see the true speed of the RAM you need to run an unbuffered Sandra test which reflects the timings as well as fsb. Unbuffered is what games and 99% of apps use 99% of the time. The true strengths of DDR434 come where you can run these speeds but WITH tightest timings of 2/5/2/2/1T. I've had alot of experience with this RAM and I must say it's the only RAM i've ever had to DDR480 without slackening the timings. A normal buffered benchmark like the one run in this review only really reflects fsb speeds, so the bandwidth will show little to no increase when tightening the timings up.
Secondly it's a known fact that the i845PE chipset based mobo's can't handle 2 sticks of RAM at 3:4 divider 99% of the time. You will only achieve the rated speed of the RAM or over when using a single stick. This is very common with the IT7MAX2.
Dave

 

Interesting dude, thanks for the info.
I seem to be really lucky with my setup as my P4 hits 3.3GHz on air and I have no probs with 2 sticks of 434 RAM, the timings are set to 2-5-2-2 also.
Can you elaborate on the bit about the buffering and the lack of reported increase in bandwidth? It didn't quite make sense.
Cheers
ATH

 

Right, a buffered benchmark is the default Sandra RAM benchmark. As you increase the fsb you will see an increase in bandwidth. Also buffered scores are far higher than unbuffered ones and this very rarely happens in real world apps. This is from my granite bay system (buffered) :



Here is an unbuffered score for my GB setup :



These are the settings you need to use for the unbuffered test :


For comparison here is my IT7MAX2v2.0 rig, both run 3.06's so the FSBs are quite low.




Both rigs using XMS3500 at the time those shots were taken with 2/5/2/2/1T timings.