Interesting Results Testing Robson Intel Turbo Memory on F3Sv-a1

I ran some more tests with a customized g2s-a1. has 4gb ram and 160gb 7200rpm sata300 drive.

With Robson:
PCmark-5676
CPU-5235
Ram-4395
Graphics-5511
HDD-5642

XP startup-14.131mb/s
application loading-8.018mb/s
general Usage-8.953mb/s
Virus scan-56.096mb/s
File Write-41.356mb/s

Without Robson:
PCmark-5475
CPU-5235
Ram-4411
Graphics-5526
HDD-4308

XP startup-8.782mb/s
application loading-5.841mb/s
general Usage-5.117mb/s
Virus scan-55.906mb/s
File Write-41.632mb/s

 

Hey Eddie, do i void the warranty if i install a robson card in an Asus ensemble (like F3Sv)?

 

highly unlikely becuase there are no warranty seals however you should not send in parts to warranty that were not purchased with the system. there is a chance you may not get back the same system or even the same parts

 

Is it just me or do these scores seem rounded off?

 

sorta rounded off, but i think the point it makes is still valid. there arent enough comprehensive reviews out there supporting or disproving turbomemory, but i'm almost ready to spend 27 bucks to test it out, and if not, then its a lesson well learned that i can pass on.

 

They are averaged then rounded off over 3 loops with and without robson. 6 different times testing.

 

Actually this is a good idea. Without the actual data, we cannot perform a statistical test to determine difference in means (testing the null hypothesis that the means are the same/diff)

 

Ouch, statistics is coming back to me.

Which test of significance would be appropriate... matched pair t-test?

Get the paired robson&no robson data for multiple notebook models, then use those data sets to conduct the t-test for each variable(pcmark, 3dmark, bootup time, et al).

 

this is a really interesting topic, i ordered a F3SV-A1, and can someone point me in the direction of a video or something that isnt bias? i seen one that was made by intel themselves (but that was probably bias) and it seemed that robson was well worth the money, and these tests look like it was too (after i read them right, i thought they were in seconds, not mb/s lol, thought robson doubled loading times ) but i would really like to see a video to see how loading times and stuff are affected, numbers can only mean so much to a person.

 

I am goinjg to repeat it for the 100th times for those who dont read. You can not see robson benefit on a bran new notebook that already boots in under 30 sec. The test were conducted in 3 pairs. 3 times with robson. 3 times without. the 3 times with robson the tests came back evenly matched. the 3 times without same story. If numbers dont mean much to a person then why is this whole forum hung up on benchmarks. Like i said before. just becuase cnet and toms couldnt figure it out does not mean that i cant either. it only means that most reviewers dont know what the heck they are talking about

 

i dunno, i read the whole post in here and i dont think you said that, i actually thought you were defending robson, but lets not get into an arguement (you would most likely win )

 

Eddie, just post a video of a laptop (new or used, SR has been out for a couple months, I'm sure you could find something) with Robson booting faster than one without. Then we will all be sated.

 

Assuming Vista needs to load around 800MB when starting (probably more), with Robson, that would take 57 seconds, and without Robson, that would take 1 minute 31 seconds. That's a huge difference. Do you actually see these differences?

 

Robson is enabled by software. It only starts to take effect once the itel turbo memory software is loaded not before. Considering that most processes are loaded once you actually see the welcome screen and after i dont expect robson to have any effect on a new machine. i am mostly interested in the added benfit of general usage. for some who experience a 2 min boot up time without robson the module would definitely make a difference. Like i said above. for a forum that is hung up on 3dmark and other benchmarks the figures i posted speak alot for the module. mostly they say that there is a very noticeable added benefit. Where you see it i dont care. Find out your self. My job is done. i prooved wrong all the reviewers out there that claimed robson had no added benefit simply by using pcmark with a recent patch that adds support for vista and by figuring out how to work the finiky driver which is not easy to figure out if you have only 1 hour to work it. i spent many days working on this stuff unlike toms or whoever else reviewed it..

 

Synthetic benchmarks are absolutely useless. They don't give a good representation of real world situations. Say for example general usage is surfing the web, typing in Word, and maybe using Photoshop. Do you see an increase in loadup times of these software? For me, I think that the only good Robson can do is to extend the battery life. But you haven't done a single test to show this.

 

Synthetic benchmarks useless? i find that strange becuase every single board memeber here who is also a gamer lives by the 3dmark score. Synthetic benchmarks are very useful as long as they work.

 

id like to see that too, any chance you could run a test on battery life?

 

I would think exactly the opposite. We like to see 3dMark scores as a general scale, but they are very much taken with a grain of salt when we are talking about real world situations. Read any thread about 3DMark and see how long it takes people to start asking for "real world benchmarks" (Game FPS) because 3DMarks "don't mean much."
The users of this board aren't dumb, Eddie, we don't hold 3DMarks as some sort of magical beacon of truth for gaming performance. Saying that we "Live by the 3DMark score" is very insulting, and quite untrue.

 

You are contradicting the obvious. the higher the 3dmark the better. I am still trying to figure out what your are arguing or protesting. scores for robson are posted. they show the technology works very well. if you want to figure out what it specifically helps then you have to find out for your self or ask someone else do figure it out. you are all welcome.

 

You've just shown what happens when someone performs tests with the aim of demonstrating a foregone conclusion. So you've really proved nothing. Your methodology as you described it is faulty. (See my previous posts on the topic.) Because of that, when it comes to Robson, I'll take Sony's, HP's and Anandtech's word over yours any day.

 

Naw man that readyboost stuff drops the access times for things from like 10ms to like 1-2. That's why the hdd sccore is good, that's all it does, that's it. I probably will not be using it though as i don't want any more overhead than vista already gives me.

also, how come hdd manufacturers are making those hybrid drives if thatstuff is useless? Apparently seagate/hitachi/western digital agree with OP

 

Ok good luck to you. If sony, hp and anand could figure out how to even enable the architecture and figure out how to work the driver then perhaps they would have came to a different conclusion. If sony and hp had as good of a r&d team as microsoft and intel who tested the technology and both agreed that it does work then maybee sony and hp would come to a different conclusion. when sony and hp both offer robson once the kinks are worked out then you will have to come back and eat your words. in the mean time i will relax and know that i am able to do and make work what few have done.

 

So have you actually noticed any differences in performance in your everyday tasks? Or is doing benchmarks all you do everyday?

It's Intel's product that requires a Microsoft product to function. Of course they're going have benchmarks that favor Robson.

 

Robson and hybrid drives are two different things. Robson is a specialized cache that the OS must manage whereas an hybrid drive is totally transparent to the OS. But again, if you were to go look into hybrid drives you ought to get benchmarks as close as possible to real-world scenario performed under sound methodological procedures by a reputable testing group. (And yes, in my book Tom's Hardware and Anandtech are reputable.) That's because whatever theoretical advantage you should gain may be totally screwed up by bad implementation.

 

holy **** you guys reply fast

 

Well you are one of few that do their research. to the rest of you. dont buy robson. your notebook will be slower then mine

 

Sure, no matter what CPU, RAM, HD, FSB speed we have vs what you have. Your laptop will be faster than ours because you have Robson... sure... Nice logic.

 

no! robson and hybrid drives go hand and hand. Robson will be a standard feature once hybrid drives are mainstream. Robson was developed as a companion to a hybrid drive sort of speak. anand and toms are reputable? riiiiight. becuase the people there spend nearly as much time working with this stuff as i did. if they were reputable both would have mentioned that pcmark did show very interesting results but unfortunately neither could figure out how to get it working. very reputable indeed. Toms reviews of asus laptops in the past have been unreliable and off balance at best. They like to pin a dtr against a mobile and point out all the negatives. thats like putting a shelby antique vs a honda civic and pointing out all the faults of the civic.

 

Regardless of what the scores say, Robson is useless if it doesn't show in any obvious way to the consumer. If you sell someone a 40 dollar piece of equipment which quadruples drive read/write speeds (or whatever) but the computer behaves exactly the same way (or not noticeably different either way) with or without it, then you basically robbed them of 40 bucks. You tried to sell it to begin with for 90 bucks, what the hell? Anyway, you say you won't notice the difference on a new system...well...you are selling people NEW systems, so basically you are asking people to buy equipment from you that you already admitted they won't even notice is in the system. So far you have done nothing to prove that it really has any noticeable effects. That bothers me.
You give us these numbers (which very well could be fake), and refuse to back it up with either video evidence or any other tests people request of you, then you get very hostile when we ask you to give us any more proof OR your complete data set and working conditions so that we can retest them ourselves...how am I supposed to think you aren't being at least a little bit dishonest about all of this? You help people around here a bit, and you are a friend of the forums, so I don't want to think that any of it isn't on the level, but the way you are acting and the feebleness of the evidence you provide forces me to think differently. You are trying to disprove a popular sentiment with basically two numbers, and then your only course of action after being put on the spot for better something more substantial is to use verbal attacks and try to make whoever is arguing with you look stupid because you "work on these machines all day" and for some reason that is supposed to make your word the gospel. The straw that really broke the camel's back for me was, again, how you ADMIT:

and on your website, advertised on a NEW MACHINE:

What else can I say to make myself more clear to you Eddie. Just because you say that you have proven it doesn't mean that you have. That is like going into court and saying "Here is one piece of evidence that implies that I am right, the case is closed and I win because I say I have carried the burden of proof" then the judge says he wants something more to make a valid judgment, as rarely a case is ever open and shut by one piece of circumstantial evidence, and you say "What part of 'I ALREADY PROVED IT' don't you understand judge, case closed. If you want more evidence get it yourself." Not only would you lose the case, you would probably get arrested for your behavior. I mean, we are your potential customers Eddie, why wouldn't you want to do everything in your power to prove to us that you are being honest? That is like throwing money away. The sad and simple answer is that you ARE NOT being entirely honest with us. I believe I speak for a lot of people on these boards when I say: until you prove me wrong, I will be not be shopping at your store, and will be wary of any advice you give."
Seriously, Eddie, PLEASE prove me wrong. I really don't want to think of you as a dishonest guy. PLEASE.

 

http://www.tomshardware.com/2007/01/31/windows-vista-superfetch-and-readyboostanalyzed/page5.html

 

My logic actually has nothing to do with it. My exprience with asus laptops does. nobody here has more.

 

another
http://www.anandtech.com/showdoc.aspx?i=3009&p=7

 

It is time to point out what I've pointed out already here. The outer quote report my own words, the inner quotes are from the very same Anandtech article:

 

that review above is poinless. all it is saying is what i said in the original post. it was very hard for me to find a benchmark that accurately measured improvement with robson present. opening 91 images in photoshop and measuring the time measures mostly how powerfull the ram and cpu are becuase that is mostly where photoshop resides. I dont see my hdd light going crazy when i am rendering an image in photoshop. Why anand picked this particular test is beyond me. Possibly a better test would be rendering a 5gb hd video file in adobe premier or vegas. speaking of which. i do that all the time. that would be a good test for me to run on one of these robson notebooks. some of the video files i render take 10 hours to complete. if robson shaves even 1 hour of the top i call that excellent results.

 

for someone who belives in benchmarks posted by anand and toms you sure are resistant to the fact that neither anand or toms analyzed pcmark scores. Why dont you go back to anand and toms who are in your book experts and ask them to analyze what it means to have a 50% and higher increase in transfer rate in 3 of the 5 tests in pcmark?

 

If it's so hard to find a benchmark that showed improvements, then wouldn't that conclude that it has no/little improvents?

Now if it does take off an hour, then Robson will have proved its usefulness. Hopefully, you'll do this test so that we can all see the 'benefits' of Robson.

 

No the reason it was hard was becuase most of the test suites do not have the array and variety of pcmark hd tests. conclusion is pcmark tests are thorough and cover alot more ground then most other benchmarks in terms of hd performance. mind you that only 3 of 5 pcmark tests had any type of results worth mentioning. if 3 of 5 tests showed 50% and higher improvement with robson while other tests dont show any that leads me to conlude that other benchmarks dont work or are concentrating on testing areas where robson is not beneficial.

 

what's wrong witht hte stuff i posted lol? It clearly shows that w/ the robson it's faster ;;;;;;;;

 

nothing wrong with it. i just think that test does not show its full potential. there are other tests that can show even greater benefit. someone will figure it out. i am not spending any more time on this for now

 

Im still doing research on this turbo memory readydrive readyboost stuff, but if its $40 on a $2000 whats the big deal? Even if it shows slight improovement now, with vista sp1 and bios upgrades wont it have even more potential in the future? I notice this board is quite argumentative and like i said im barely starting to research this so im willing to accept what i just said to be wrong, can anyone sum up what turbo memory, readydrive and ready boost are? Also, something that confused me, on intels site it says that this turbo memory is for vista ready computers, yet i saw scores for XP loading times being faster with robson on this board. How comes XP is using it?

 

I feel the same way. But I am also new so "I may be wrong as well but This is what I think :> Cheap for a little boost in performance.

 

Also, why did anand not post results for Turbo Memory Enabled on a 2gb machine. Is turbo memory just a replacement for another 1gb ram? Turbo memory enable on a 1gb system was practically identical to turbo memory disabled on a 2gb system? Is the photoshop test they are running to load images not the best way to demonstrate turbo memory's advantages?

 

the turbo memory is supposed to alleviate the negative effects of HDD seek times. It caches stuff that's already on the hdd so that when the OS needs it, it can pull it instantly off the flash instead of having to wait for the hdd to find the data on the disc platters.

HDDs have more speed after it's actually reading the data on the drive however, so it's good to have the OS read from the flash while the hdd is seeking, and then switch back to the HDD when ti's ready to feed the data to the OS

 

In that case, is there any benefit of upgrading the HDD to 7200rpm from 5400? Would a 5400rpm with turbo memory work out the same as a hdd running at 7200rpm?

 

If you have a ton of ram, the 7200 rpms should help more than the turbo memory, If you have limited ram, then the turbo memory will help more than the 5400rpm drive.

but brand/model on HDDs matters more than rpm imo. I've seen several benchmarks where a 5400 seagate has outperformed a 7200 hitachi/western digital *consistently*

edit: with the same capacity.

 

Im getting a laptop with a 160gb seagate 7200 rpm HDD, 2 gb (2x1gb dual channel 667mhz), and was thinking of just getting this turbo memory (1gb) for an extra £18.

 

Ive heard getting 800mhz ram for santa rosa platforms is pointless as it only uses up to 667mhz, is this true?

 

It will help, you won't notice it though lol, not on a machine like that

edit: about the turbo ram lol

 

The ram on santa rosa platforms only clocks up to 667Mhz (not on par with the 800Mhz FSB).
However, 800Mhz ram often has better latency, so there will be a slight difference though it won't be really noticeable. It's not worth getting

 

Okay, this is the deal.
Robson is a code name for a Flash Memory module attached directly to the motherboard via a mini-PCI slot. It uses ReadyBoost in exactly the same way a 2GB SD Card would, except it is NAND memory and has a direct interface, so it should work a lot faster than a regular readyboost memory module. It is not some wild new technology, it is just assembling two technologies we already know a bit about, NAND memory and the Vista ReadyBoost feature and adding an interface with the HDD to take away the lag that happens as the drive speeds up and slows down searching for data. The problem with compatibility comes in when Vista tries to address the Robson, but doesn't know what to load onto it. According to Toshiba and Sony, Vista doesn't realise what the Robson module is there for, so it goes about ReadyBoost in the usual way without taking advantage of the speed and its relationship to the hard drive.

Robson memory is a great idea, but the execution has been bad so far. On an ultraportable system that uses a 1.8" 4800 RPM drive, Robson could show massive increases in performance (or at least noticeable ones). Especially since 1GB Robson is cheaper than a 1GB stick of the extra small RAM that is used in those computers. The way it is supposed to improve battery life is that it alleviates the stress from the HDD allowing it to run less constantly, without having to read and write every bit of program data that is being transfered. However, in its current state it doesn't work that way (a Vista limitation), since the memory doesn't know what it should and shouldn't be storing at any moment everything is still writing to the HDD (instead of part to the Robson and part to the HDD), thus the drive is running just as much with or without.

Regardless of all of that, even if Robson worked exactly the way it should, the results on a machine like the F3 would be minimal. You would see possibly 3 to 4 seconds of boot time difference and probably the same for load times in games. Robson is to make up for bottlenecked parts of mobile hardware. When you have a T7500, a 7200 RPM HDD (close to 3Gb/s data rate) and 2+ GB of 667 speed RAM, you don't really have any components that are slower than NAND (even though it transfers faster than the HDD, the rest of the system can only process about how much the HDD is sending, so they are essentially equal by all practical terms). The only thing it stands to benefit is battery life, but as I said, poor implementation has thus killed that plus. Hopefully about the time Robson 2.0 comes around MS will be ready to release an SP and the other hardware will be craving the speed boost, I am hoping that I will eventually be able to get a 4GB Robson module to use as a page file.

I think that is absolutely everything I know about Robson. Anything else is either above my head or just haven't heard it. I guess that is really all I have left to contribute. Hope that helps someone!