chispy Posted May 6, 2021 Posted May 6, 2021 Good afternoon gentleman. I just found out 48 hours ago about as ssd 2.0 getting global points wish is awesome as we really needed it a 100% storage benchmark ? , but here comes the but ... I trully believe that the rules by hwbot need an update for this bench as it is too easy to use software ram disk , cache booster and what not to boost the scores on your storage with dodgy software. I have already seen a couple of scores using primo cache ram cache software and other software crap , but it is easy to detect some of those out of the ordinary scores , while others are more difficult to detect. In light of the Shenaningans going on with this bench i propose for the rules to be updated with a picture of your hardware set up and storage use clearly showing the storage used ( nvme / ssds , Raid Cards , etc... ). Also it would be very helpful to show thru device manager under disk drives the page where it shows exact details of the disk drive storage used + same place in device manager also show Storage controller. I believe it is nessesary to maintain the integrity and fair playground with correct scores while submitting the results. Here is a good example of what i would like to see for the sake of fairness , otherwise this benchmark will become pointless to run : @Leeghoofd 1 1 Quote
Crew Leeghoofd Posted May 6, 2021 Crew Posted May 6, 2021 Good will add the disk setup picture straight away to the rules Angelo 1 Quote
Mr.Scott Posted May 6, 2021 Posted May 6, 2021 You know that's not gonna help. Pic's are and easy fraud. Dump that bench. It will be nothing but trouble. 1 Quote
yosarianilives Posted May 6, 2021 Posted May 6, 2021 34 minutes ago, Mr.Scott said: You know that's not gonna help. Pic's are and easy fraud. Dump that bench. It will be nothing but trouble. Could be said for a lot of benches, anything without file tbh... I think the idea of a storage bench being worth doing is a good one but hard to implement. Maybe @_mat_could make a crystal disk bench wrapper or similar free bench for BM? No idea how much work that would be but probably the long term viable strategy for storage bench globals. In the meantime I think requiring additional proof in a screenshot of an existing bench is a good idea 1 Quote
chispy Posted May 7, 2021 Author Posted May 7, 2021 4 hours ago, Leeghoofd said: Good will add the disk setup picture straight away to the rules Angelo Thank you Alby , appreciate it. Quote
Crew Leeghoofd Posted May 7, 2021 Crew Posted May 7, 2021 Either the community makes it or breaks it... Bugged stuff will be removed and users warned and/or penalized. 3 Quote
TASOS Posted May 9, 2021 Posted May 9, 2021 Are we really sure about this benchmark ... ? Is it tested enough ? Does it produce consistent results or are we playing lottery once more ? I get different results each time i run it ... and i'm also getting big difference in scoring when running each subtest alone 1 Quote
Crew Leeghoofd Posted May 9, 2021 Crew Posted May 9, 2021 Storage benchmarks are hardly stable in output alike eg Superpi... this might be one of those run re-run benchmarks... 2nd run is always faster here on my daily... 3rd run slower... its widely used by review sites, so can't be that bad right ? 1 Quote
TASOS Posted May 10, 2021 Posted May 10, 2021 I did 3 runs within a 10min period with a single Samsung 860 Evo on a Sata2 controler. 1st run scored 1611 2nd scored 1962 3rd scored 3071 ? Which one should i submit ? 1 Quote
yosarianilives Posted May 10, 2021 Posted May 10, 2021 11 minutes ago, TASOS said: I did 3 runs within a 10min period with a single Samsung 860 Evo on a Sata2 controler. 1st run scored 1611 2nd scored 1962 3rd scored 3071 ? Which one should i submit ? Did you install the Samsung sata ssd driver? Quote
Crew Leeghoofd Posted May 10, 2021 Crew Posted May 10, 2021 And that is without any caching software ? Hard to believe... 1 Quote
yosarianilives Posted May 10, 2021 Posted May 10, 2021 4 minutes ago, Leeghoofd said: And that is without any caching software ? Hard to believe... If he installed Samsung driver he has ram cache and doesn't realize it... 1 Quote
TASOS Posted May 11, 2021 Posted May 11, 2021 13 hours ago, yosarianilives said: Did you install the Samsung sata ssd driver? 10 hours ago, Leeghoofd said: And that is without any caching software ? Hard to believe... These tests were performed on a office pc , that is used for logistics and internet. I have no clue if there is any kind of extra software installed , but i will check next time. Quote
moi_kot_lybit_moloko Posted May 12, 2021 Posted May 12, 2021 can i bench old hdd such as IBM DTLA/Seagate Medallion/Quantum Fireball on this benchmark? 1 Quote
moi_kot_lybit_moloko Posted May 12, 2021 Posted May 12, 2021 On 5/10/2021 at 7:12 PM, TASOS said: I did 3 runs within a 10min period with a single Samsung 860 Evo on a Sata2 controler. 1st run scored 1611 2nd scored 1962 3rd scored 3071 ? Which one should i submit ? third, sir 1 Quote
Matsglobetrotter Posted May 14, 2021 Posted May 14, 2021 I am not convinced the AS SSD is buggered but who knows. Meanwhile i know the SSD drives in general are buggered. Seems that most SSDs are so hot they are always on the limit so always swap down the speed unless properly cooled. If i however run a properly cooled SSD like the optane drives i have, the scores are pretty consistant as they seem to manage to keep up the transfer throughput speeds over longer time than an ordinary Samsung drive for example. This is same behaviour i see on an ordinary file transfer in windows. a normal ssd start out great and then usually takes forever anyway once it heats up a bit. In any case i do like the idea of having an SSD bencher in the bot with scores. Especially when we can mix in raid configurations. meanwhile windows really takes a toll with all its crapware on the drive throughput so a bit of crap removal really do help ? 1 Quote
Matsglobetrotter Posted May 14, 2021 Posted May 14, 2021 18 minutes ago, Matsglobetrotter said: I am not convinced the AS SSD is buggered but who knows. Meanwhile i know the SSD drives in general are buggered. Seems that most SSDs are so hot they are always on the limit so always swap down the speed unless properly cooled. If i however run a properly cooled SSD like the optane drives i have, the scores are pretty consistant as they seem to manage to keep up the transfer throughput speeds over longer time than an ordinary Samsung drive for example. This is same behaviour i see on an ordinary file transfer in windows. a normal ssd start out great and then usually takes forever anyway once it heats up a bit. In any case i do like the idea of having an SSD bencher in the bot with scores. Especially when we can mix in raid configurations. meanwhile windows really takes a toll with all its crapware on the drive throughput so a bit of crap removal really do help ? hmmm actually i retract my statement above on as ssd hehe. I have run it many times last month. However just tested a new thing. put priority to high then it fails to give correct results randomly. The score on the left is what normally should be seen. the score on the right is not possible on the read 4K-64Thrd. plain vanilla Win10 enterprize with no cache active. 1 Quote
chispy Posted May 14, 2021 Author Posted May 14, 2021 I have been playing and testing different things on this bench , sometimes you get a one off with a high score and i think is the cache on the nvme drive or windows well been windows. I really like this bench as finally we have a pure storage bench that it was needed here. @Matsglobetrotter you found a new tweak ? , thanks for the heads up , now we know where to look for a bugged run with a high score. Quote
TASOS Posted May 14, 2021 Posted May 14, 2021 Start running each subtest alone ... and re-run and re-run , and you will see miracles happening * and , yes i confirmed that the pc i tested had Samsung magician installed , with rapid mode enabled. Quote
yosarianilives Posted May 15, 2021 Posted May 15, 2021 (edited) 5 hours ago, TASOS said: Start running each subtest alone ... and re-run and re-run , and you will see miracles happening * and , yes i confirmed that the pc i tested had Samsung magician installed , with rapid mode enabled. Samsung magician is ram cache, and guess what type of software is banned Edited May 15, 2021 by yosarianilives Added rules 1 Quote
chispy Posted May 15, 2021 Author Posted May 15, 2021 (edited) Can you guys spot what's wrong with this score ? take a guess ... https://hwbot.org/submission/4746543_godsgamer_as_ssd2.0_ssd_18998_marks Yep , that's correct , cache ram software " primo Cache Ram cache Software , even the icon is on his desktop ? . Scores like this are easy to spot , 4k random read /write is out of this world with cache ram software. It should not be that hard for anyone to know this and spot , out of the ordinary runs. This is the second time this user has submmited scores like this and was reported @Leeghoofd , first one was 17k + and now 18k + with primo cache ram software. . Edited May 15, 2021 by chispy 1 Quote
Matsglobetrotter Posted May 15, 2021 Posted May 15, 2021 meanwhile I think any AS SSD submission that has 4K -64thrd higher than the sequential speed on either read or write must be buggered. It can be close to sequential but never above without something else being at play as in cache. Interestingly the above score also have a latency of 0.006ms. There is no SSD that has such a latency thus it indicates cache in memory. 1 Quote
TASOS Posted May 15, 2021 Posted May 15, 2021 5 hours ago, yosarianilives said: Samsung magician is ram cache, and guess what type of software is banned You dont get the meaning of my posts. Perhaps i dont express myself correctly. I know very well what is allowed and what's not. Dont focus to score as numbers (score is out of line cause of the Samsung rapid mode). Focus on the variation of scoring that i am reporting. Quote
TASOS Posted May 15, 2021 Posted May 15, 2021 3 hours ago, chispy said: Can you guys spot what's wrong with this score ? take a guess ... Becides all other already mentioned. Scoring of first line "Seq" ... is not centered and there is also missing the MB/s 1 Quote
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.