Interesting Results Testing Robson Intel Turbo Memory on F3Sv-a1

So as the title reads I have been messing around with a 1gb robson chip using vistas readydrive technology. I picked the f3sv as the test platform. Testing took me almost a week because quite frankly i could not find any benefit to robson using a half dozen benchmarks. However my last attempt using pcmark05 (vista patch) showed some interesting results. Now please keep in mind that results can and will vary depending on the actual config but this is what i found on a fairly basic f3sv-a1. I ran 3 loops of pcmark (full version all tests with robson and 3 loops without. Here are the averages

With Robson:
PCmark-4450
CPU-4890
Ram-4150
Graphics-3100
HDD-4770

Without Robson:
PCmark-4490
CPU-4950
Ram-4150
Graphics-3150
HDD-3720


After Bios Flash:

With Robson:
PCmark-4800
CPU-4980
Ram-4150
Graphics-3400
HDD-4600

XP startup-11.067mb/s
application loading-6.349mb/s
general Usage-6.210mb/s
Virus scan-55.752mb/s
File Write-35.981mb/s

Without Robson:
PCmark-4500
CPU-5000
Ram-4150
Graphics-3400
HDD-3600

XP startup-6.773mb/s
application loading-4.829mb/s
general Usage-4.017mb/s
Virus scan-56.607mb/s
File Write-35.057mb/s

Now to someone who is looking for a needle in a haystack they may say that "it performs better in the top 4". to me the scores above say the top 4 are equal. Now after testing and retesting and retesting it is evident that the series of tests in the HDD loop are obviously higher when a robson chip is installed. By about 25%. I will edit this post in a few days and show 2 pics of the individual hdd scores to better explain what is going on and specifically where vista benefits from readydrive. Also i though it was worth mentioning that with robson installed the system halted shortly after the asus logo splash 1 in 5 boots. This will be addressed by a bios update no doubt sometime in the future.
On a totally separate note i wanted to mention that every single review i have seen out there by a veriety of different sites and reviewers claimed that the technology is flawed and does not help. That may very well be possible however the f3sv from asus definitely shows otherwise.

 

Thanks for the information Eddie. Is there any chance you could do a battery life test? Intel claimed better battery life when they were introducing the technology.

 

If possible, could you also test boot times?

 

the detailed hdd tests iam working on will reveal avg boot times and such, i did not notice immediate benefit in boot times. The battery life is about the same as it was in the napa platform. I did not notice better or worse life. expect about 2-3hrs out of an 8 cell. Again it really depends on the user, the os, and the process.

 

Interesting tests. +rep!

 

That's quite a bump in HD scores...

 

looks like i have to retest. just installed the lates bios and now all the scores are higher.

 

Can we get some error bars? If they are equivalent scores then the error bars will overlap. In other words, can we see ALL of the results, so we can take the low and the high score for each and see if they are equivalent by our own standards?

 

Eddie, given that you tried 6 other benchmarks and didn't see any results, do you really think this one case is statistically significant?

Is each of those numbers just one run, or did you make several runs with and without and average the results for each?

Really appreciate your hard work trying to get to the bottom of this, but one improvement out of 6 benchmarks isn't too convincing.

 

Are the results the same when you use a usb or an external card?

 

That's a very significant problem in methodology. 5 benchmarks are thrown out because they don't show the desired results and 1 benchmark is kept because it does? That would make a statistician cry.

 

3 runs with robson 3 runs without. 3 runs with similar results to each other. 3 runs also with similar results to each other. the only difference was the hdd score. hdd score was about the same with robson, and also the same with out robson within the 3. that is pretty significant to me. now i have to redoo the tests because flashing the bios increases all the other results while keeping the hdd scores unchanged. the only real change is the hdd score from robson to without robson. The tests are accurate and show consistency all the way. the only problems are the booting issues and getting the module to work right, it almost functions like a donor organ. if you take it out or replace it, then it will not work right the first few hours of use. I am still trying to figure it out. The only thing evident is that the hdd scores are increased with robson by a predicatable ammount and decreased by the same.

 

That is exactly what i did NOT say. If you read above the averages are out of 3 very similar benchmarks. 3 improvement in 3 benchmarks. 3 reductions in 3 benchmarks. I would have to be a moron to think that one improvement in 6 benchs was indicative of anything. been doing this for many years guys. Give me some credit.

 

What the heck does this mean then?

 

Save your money

 

It means i could not find any benefit using any benchmark OTHER THEN pcmark. This means that based on pc mark results robson is well worth it becuase what intel promised is true based on hdd pcmark scores. Am i going crazy or are people not reading the scores correctly? How about reading the entire post not just the first sentence

 

Yep. In one word, RAM. Whatever you are spending on Robson, invest in RAM.

 

That is brilliant. Its almost as good telling someone "whatever you are spending on your new brakepads invest in better transmission". What does robson have to do with your system ram? absolutely nothing. one technology enables the other. robson module is cache for your hardrive.

 

Well the problem I have is that if it does what Intel says it does, there should be improvements in boot times, hibernation times, etc. The HD benchmark in PCMark05 is one specific test, and might not even translate into any real-world improvement in productivity.

For example, both HP and Sony have accused Intel of choosing an unrealistic internal benchmark - opening a ton of Photoshop documents faster then any normal user ever would.

Based on the way Vista uses system RAM to cache stuff, I think the money would be better spent on RAM.

But if the reliability isn't there, all of it is a moot point anyway.

 

the hdd suite of tests is not one test it is actually 8 or so. i am still working on the screen shots. all aspects of the tests show improvement

 

RAM is more cost effective performance improvement than Robson memory is. It is not comparable to brakepads, it won't destroy your system to not have Robson memory. It is more like saying "whatever you are spending on a spoiler invest in better transmission."

EDIT: And again Eddie, to take your statistical feet out of the fire, publish ALL of your results. Period, no exceptions, then people will lay off if they are what you say they are.

 

what does ram score have to do with this? ram is not the issue here neither is the cpu or the video. read the scores above. ram, video, gpu are unaffected by robson. ONLy the hdd scores are affected by it. The scores are already published. Excuse me but my credibility is not of fire here. The scores are for people to look at and analyze. i will correct anyone who analyzed it wrong. They are what i say they are that is the end of the story perion no exceptions. Done deal.

 

Robson, in it's current implementation, is supposed to be work with vista's Ready Boost. So it's perfectly valid if I say that RAM is worth the money, compared to Robson. It doesn't improve boot times, it doesn't load programs faster, it doesn't lead to a faster sleep/Hibernate. It does nothing it originally was being touted to do.
I stand behind my statement.

 

I think the main arguement here is if you're buying a new Santa Rosa laptop right now, you will get more of an overall performance boost by spending an extra $50-100 on RAM then you will on spending the same money on Turbo Memory. I agree they are different technologies, but right now Intel Turbo Memory does not seem to provide a good cost/benefit ratio (that could change with better drivers--but there is no assurance).

 

ok here are the final test scores with new bios. the bug with startup is related to the intel turbo memory concole. when the pc is rebooted via concole the first time it halts during post. I guess its either intels problem or the bios.

With Robson:
PCmark-4800
CPU-4980
Ram-4150
Graphics-3400
HDD-4600

Without Robson:
PCmark-4500
CPU-5000
Ram-4150
Graphics-3400
HDD-3600

 

Sir...are you blind. I JUST SAID it does do exactly what it was meant to do! One more time! Look at the hdd scores. memory is always worth the upgrade. you are in the wrong thread. this is not a thread about ram. this is a thread about robson and its inherent ability to boost your systems performance specifically with read/write speeds which do include all of what you just said .

With Robson:
PCmark-4800
CPU-4980
Ram-4150
Graphics-3400
HDD-4600

Without Robson:
PCmark-4500
CPU-5000
Ram-4150
Graphics-3400
HDD-3600

 

And it isn't that RAM is on trial here, we are saying that performance gain/price ratio is much higher for RAM than it is for Robson. And when I say publish ALL of your results, I mean the results of every single test, so we can do the averages, error bars, K-Values, etc. Not just the average results that you got from your tests. By resisting you are hurting your own credibility.

 

To someone who has 50$ to upgrade their system that may be the case if it lacks ram. to someone who purchased a notebook and has ample ram in it.....Not everyone here is broke. alot of people have and want to spend money on this technology if it works. If i were buying a g1s for example. I would not even bother adding anything but a faster hard drive and a robson module. i would pick a faster hdd and robson over an upgrade from 2gb to 4gb and so should anyone else unless they have a SPECIFIC need to have more then 2gb of ram.

 

or you can get a faster HDD and 3GB of RAM.

 

Ok how about anyone here that is "deaf, or blind" please stop posting. anyone here that is talking about oranges not apples please stop posting. ask relative questions to robson or i will simply ingnore you. Alex like i said above detailed hdd scores (there are 5 of them) will be posted shortly.

 

I am afraid that I am mising something, but there is not a single boot time score here. Neither are there Hibernate/Sleep, and resume times. I see meagre improvements in some synthetic benchmarks. Exactly why I went with 4GB RAM instead of a robson module.

EDIT:I read the previous post after I posted this, so discount my RAM comment. But to rephrase what I said, does Robson improve boot times?

 

a correction to someone who brought it up above. robson does not use vista readyboost. its part of vista readydrive. there is a difference but both technologies basically aim to do the same. provide faster read/write speeds.

 

Ok here are the final 5 scores from the hdd tests (5 tests total).

With Robson:
XP startup-11.067mb/s
application loading-6.349mb/s
general Usage-6.210mb/s
Virus scan-55.752mb/s
File Write-35.981mb/s

without robson:
XP startup-6.773mb/s
application loading-4.829mb/s
general Usage-4.017mb/s
Virus scan-56.607mb/s
File Write-35.057mb/s

Ergo. XP startup tests are almost double mb/s. app loading are noticeably higher. general usage also noticeably higher. no change in virus scan. no change in file write. If the suite had more then 5 tests i am sure results would provide more insight. The end

 

But Vista is the OS that Robson is designed for...why not use that?

 

f3sv comes with vista home pre. that is exactly what i am using! results are from PCmark05 with vista patch from future mark. the same company that makes 3dmark.

 

What is that about then?

 

xp startup tests are tests run by pcmark. so are the rest of them

 

I haven't PCMarked a computer in Vista before, so I can't argue with you, but could you post some times? That is what the consumer wants to see. Robson could double the read/write rate in every category, but if it doesn't know what to read or write (which is what I have heard is the problem), then there still would be no noticeable performance increase to the lay user, who is (in the end) the one buying it.

 

the scores above speak volumes. there is no need for further testing robson has a very noticeable result in boot times and general usage. i can sit here with a clock all day long and time stuff. that is why there is a pcmark. it removes possibility of human error. there is no error in those scores. what intel claims to be true is infact true according to pcmark05.

 

Great work!
I truly see the benefit of Robson in hdd performance and really believe in it.

Which Asus notebooks or other notebooks have Robson Intel Turbo Memory at the moment? Unfortunately, F3Sv is not available in HK. I am only a light gamer. So G1s might be too much for me. Besides, it does not have Robson technology implemented, right?

 

the main reason that robson is not integrated in to any brands now is becuase there are inherent driver problems. at the least the driver is problematic. I see the benefit too however bios and driver issues need to be worked out before manufs start integrating. most asus models will support the robson module or a half height asus module. so far a8sc does not. g1s does support the itm module but needs slight modification and prior exprience to put it in.

 

I ordered Robson for my Compal IFL90. For $40 I can't go wrong.

So the way I understand Robson technology is that it "learns" as it goes. Is this correct?
Tim

 

I wouldnt say it learns as it goes. I think the problems i had mostly related to the intel turbo memory concole and driver. Retesting in 3dmark did not give better results after every test. You should have confirmed to make sure the compal unit you ordered takes a full size robson. some of the manufs out there are making it so you have to buy a proprietory module from them. asus s96s and g1s (works with full size after a mod) are two models from asus. a8sc doesnt even have a visible port to put one in

 

I'd like to see a before and after of a 512mb SD card. It may not be the exact same thing as Robson, but Sony is claiming it's a better price/performance ratio.
I'm excited to get my G1 in a couple of days, and I may just put a 512 sd card in there and leave it.

 

Ah I see. So basically it all comes down to drivers to determine performance.

Well I got 1GB Intel® Turbo Memory (Robson) so I think that should be full sized robson.
Tim

 

I have no knowledge about this technology, but is turboboost the same or similiar to vista readyboost? So for example with the asus c90 I want to buy a usb readyboost ram chip for 20-30 bucks so will that give similar results to to the robson scores u got?

 

eddie, have you tried running hdtune?

 

I have not tested readyboost mainly because the few sd cards i have laying around were incompatible (too slow i guess). You cant just use any stick. has to be fast enough. I dont see how sony can claim this. Readydrive is inherently faster only becuase any sd card or usb stick have to go past the usb host controller. Not too sure what the internal mini pci-e ports use as their controller but i am sure the intel schematics for the 965 chipset explain it. I think the benefit behind robson is its direct link between the hdd via the pci-e port.

 

didnt try hd tune personaly but i think my tech said it didnt work with vista. too late now though. i wont get to run more tests till we get the asus half height robson

 

i used hdtune on a vista machine a couple months back. it would be nice to know if robson affects access rates....