Yamunsa
-
Posts
187 -
Joined
-
Last visited
Content Type
Profiles
Forums
Events
Blogs
Posts posted by Yamunsa
-
-
Yeah its not so good for pcm05... better 1882ix with 4GB Cache
yup, i'd be comfortable saying that i don't think an lsi 8 port controller is going to beat the areca. at best the 9265-8i can achieve 700mbs GU and 1300mbs VS. i've got some more testing to do with an expander but i have a sneaky suspicion the card won't push past these numbers in pc05.
-
just note:
This is more than just another step forward in competitive benchmarking, OCKO is an evolution of online sport.rofl
-
cheers fellas. these are still test runs without optimisation
it is worth noting that the G1 is not an efficient board and the bios i'm running currently does not have direct pcie control. vs was low on this run but unlike moose i'm running an LSI that doesn't have 4GB onboard cache lol.i'm also playing with an expander that is currently saturating a single port rather than bridging / switching / spreading the load across all ports. (new cables in the post so hopefully better results after the next round of experiments). 10 ssds listed as the bot doesn't have a larger number to chose from
-
nice run fella. that xp startup is spot on.
-
Why would you need 200GB of cache lol.
easy - upgrading sme's running wsus, kes exchange, kav and rest that are having their iops ruined from poor setup. many 'solutions' put together in r1 to cut costs coming back and biteing is just 1 example. large db usage by multiple users on a ram shy setup is another.
cachecade was a great idea but many enterprise operators have an unwarranted fear of ssds due to their failure rate, (stupid as a caching drive failure will not lose mission critical).
the 81** is simply an onboard cachecade solution with a big price hike. a great idea from lsi to comfort tech serve agents wanting to improve iop delivery, (that can be bettered by buying a lesser card, expander & ssds).
-
Thanks!
It's not stated in the rules, so I am going to ask: Is LSI's CacheCade and Fastpath software allowed? I guess it is, because it's not very different to Intel RST and AMD RAIDXpert, but I just wanted to be sure!
it's not been ruled on yet but my take on it as an LSI owner:
buying a fastpath key, hardware or software activated, will initialise the cards native hardware feature but because it's a hw level feature i'd guess it'd be aok.
cachecade is a firmware level feature so just like rst and xpert i'd expect it to be seen as aok to use again.
either way, to make things easier to aducate their use by both yourself and moderators:
in a pure ssd setup you will get better results using write back with BBU which effectively disables fastpath so it's not worth buying or using.
cachecade delivers very marginal benefits with an ssd array. an extra 2 ssds would cost less and deliver a better result. if you have 8 ssds plugged in, the best result i have managed with cachecade was 5 x r0 to bench and 3 x r0 dedicated to cachecade which gave me an extra 100mbs in virus scan over a clean 8 x r0. again it's not really worth buying over more ssds and playing with your settings
-
2nd areca? wheres the chenbro lovin
-
well it was worth a shot - RST added to the too hard basket :celebration:
-
just to be clear regarding RST - yes it has been available to install and use since 2003 known formally as Intel Matrix RAID, MSM and from 2010 RST. as today, like it's previous versions it's intended purpose is firmware level raid featuring the option to select different areas (e.g. partitions or logical volumes) on the same disk and assign them to different raid devices (e.g. ICH device).
in 2011, intel brought out ver.10.5 of RST with a new component called SRT. quoting the wiki:
Smart Response Technology (SRT) (pre-launch name SSD Caching) is a proprietary caching mechanism introduced in 2011 by Intel for their Z68 chipset (for the Sandy Bridge–series processors), which allows a SATA solid-state drive (SSD) to function as cache for a (conventional, magnetic) hard disk drive.SRT is managed by Intel Rapid Storage Technology software version 10.5 or later, and implemented in its device driver and the Z68 motherboard's firmware (option ROM). It is available only when the (integrated) disk controller is configured in RAID mode (but not AHCI or IDE modes) by implementing a style of RAID-0 striping. Write-back (Maximized mode) or write-through (Enhanced mode) caching strategy can be selected by the user. The maximum utilizable cache size on the SSD is 64 GB. Caching is done at the logical block addressing (LBA) level, not the file level.
first time around, SRT doesn't really do all that much. It's only on the second and subsequent instances when frequently accessed data is called that SRT kicks in and Windows reads directly from SSD cache, bypassing the hard disk entirely. When that triggers, the speed boost over identical data called from the hard drive can be enormous.effectively pc05 submissions made before may 2011 are completely free of scrutiny.
noting sentiment that RST (SRT) should be allowed so that competitors who do not own a 3rd party controller can compete with those that do would make pc05 the only benchmark with this degree of leniency. rules are already in place that don't allow mip to be used on a 560ti so it can compete evenly with a 680. I would go as far as saying that apart from this discussion hardware owns on hardware bot.
with SRT still being in it's infancy lord knows how it will develop and whether hwbot will need to revisit this same issue in the future, (let alone the development of ms readyboost).
understanding a press release will be put together in the next few days i look forward to clarification on the official read on how hwbot see tricking benchmarks into giving a score that’s not real fits with software caching in pcm moving forward.
-
and again 1732 vs scored with only 2 ssds with rst and 32k stripe taking wr 4 core.
-
this is a shame as Turrican was by far the most active in the support ticket section recently.
-
when used to cache on ramdrives not ssds but i agree that sw caching can be pushed to the extreme. it's a route i'd much rather see halted before it becomes abused.
here is a perfect example of a score that could fly under the radar if the user didn't list what he'd done - link.
i know i have posted hard today but imo the sw caching discussion should be had. being used to seeing things open or shut the door very much looks ajar.
-
i know i've had a lend of you before but over here i have the utmost respect for you, (even more so when concerning pcm).
i think everyone needs to understand there is no rule change here, its simply rule enforcement and in some tricky cases where things might fall between rules, clarification of rulesin responce; the most widely used rules page for pc05 is dated May 15, 2008 well before the notion of this sort of software was about, (noting ramdisk was banned outright). link.
using rst & xpert raid to trick the benchmark into giving a score that's not real, changing the perceived speed of the benchmark, is exactly what is happening. as linked above the theoretical and advertised limit of drives are being exceeded in the subtests. even these speeds do not take into account the translation of synthetic performance into pc05. 2 x retail sata lll ssd in r0 cannot physically hit 1200mbs let alone the convertion of speeds in to pc05's virus scan.
the president of permitting / accepting even condoning the use of tools that previously has been shown to effect the benchmark into giving a score that's not real is no reason to retain the acceptance of them. F2D is an excellent example. we are still in the process of moving on from that era.
-
Everyone is never going to be happy, facts of life
very true.
Intel RST has long been allowed and AMD raid xpert is its equivalent from another vendorrst and xpert raid has been around, (used in conjunction with pc05), for less than 2 years.
Other software caching programs have long been bannedspecific note to fancycache and perhaps supercache in regards to using system ram not other ssds?
You think it's a nightmare to moderate rst? Try going back through the entire pcmark results and removing all the submissions using RST? Plus all the people using it for over a year and now their results being removedSimilar to Lod this might not be the ideal situation but we think it's the best taking in all the factors
i disagree. i think handling rst / xpert raid or indeed moderating old subs for rst could be quite easy. four different methods:
(1) ask the community to report submissions and over time the problem will solve itself rather quickly as demonstrated recently with TE & TW.
(2) if hwbot has a working relationship with futuremark ask them for access or a list of results in their pc05 db with scores exceeding a hdd general usage level of 300mbs and compare the md5 hash with hwbot submissions of known users of sw caching. with db level access at hwbot this would take less than an hour.
(3) do it the hard way and scrutinise individual scores. this sounds worse than i think it would actually be - there are not many hwbot members benching pcm. even fewer who are actually competitive. realistically you are looking at the top 30 global subs for 1, 2 & 4 cores to identify those who use rst / xpert and those who have an array. moving forward the individuals best results would need to be looked at to confirm the use of sw caching and a circular pm could be sent to the member requesting self moderation. i believe this technique could be divided amongst a handful of admins and could be sorted over a weekend.
(4) draw a line under the current subs and move forward with a final ruling. this bridge has been crossed before dating back to AM3 and the consideration of existing scores.
We want a level playing field in the future without impacting existing scores in our ranking, no one said moderation would be easy, it won't, but people know now if there result is out of line they will be asked to explain it, so try it on and see what happensi agree, however the recent purge of scores highlights both the communities and individuals' backlash to the use of tools excessively altering the realistic score of the benchmark. the momentum to finally clean up pc05 is there. it is up to the community as a whole to finish the job.
Blatantly breaking hwbot rules won't just result in submission removali sincerely hope this is an indication to enforce the rules in a fair, blanket and firm fashion.
to comment overall i think the following is worth pondering:
linkThe world of overclocking and the community it’s founded on is in essence not so much different from any other field of interest or social group. We have our own language, our own set of principles, our own set of shared motives and goals and (virtual) places we visit. Just like any other social group, once in a while we are forced to face an existential crisis; a problem that pushes us to think about what overclocking and benchmarking is, or what it is supposed to be. Over the years, we’ve had lengthy (and interesting) discussions regarding the usage of extreme cooling, a debate on whether or not graphical artifacts should be tolerated, if adjusting LOD values should be considered legit and many many more. -
the current thinking is, rst/raid xpert allowed
taking this example - link : StrategosSan comments that he only used 2 x Crucial M4 SSDs and his virus scan result exceeds the phisical limit of the drives let alone the translation into pc05.
software cacheing looking to be deemed AOK.. this is a real surprise and a dissapointing ruling.
-
i think the stumbling point of the discussion is about how to enforce the rule and deal with previous submissions, that have already been made using these programs, than whether the software should be permitted.
knowing it would be much easier to allow these programs, then their use does not need to be policed, still doesn't make them right... if the use of software caching is permitted then using tools from dataplex, fancycache/supercache (pointing to ssds not system ram), velobit, enhanceIO or even using mklink to soft link directories on your SSD to directories in program files as a cache medium would have to be permitted because it would be impossible to draw a firm line in such a grey area.
from my own research i've seen it to be quite easy to spot especially when the majority of submissions with HD GU over 300mbs and VS over 600mbs more often than not have listed their storage and or have posted a pic of their setup. ease of spotting is for me is further emphasised using my own or steve ro's submissions to see what large acard, iram or ssd arrays can do.
permission of rst also limits the advantages to participants running Z68, Z77, H77 & Q77 as, as far as i'm aware it only works on those platforms. so if folks want to have a bash at older HW records they will have to have a large array only to watch it get smashed by inferior setups if they indeed move it to current gen platforms.
looking at what pro has posted previously i think the writing is on the wall for raid expert & rst and any 3rd party caching program for use with pc05. it's just too large a can of worms to be allowed and the impact on other PCM benchmarks is obvious.
highlighting notable points from the first page:
why keep pcm05?for those that are saying "get rid of pcm05", we are basically at that point and i am sure if this doesnt work then that is what will happen, but there are lots of reasons to keep this benchmark, for example,
we have guys like Steve and Aristides that are always strong in this bench and are getting beaten by tweaks that they are not comfortable using and rightly so, this benchmark has a place and is a source of enjoyment for lots of people, so i believe we should have a proper effort to try and fix it up before abandoning it
i' pretty sure steve ro has not made a submission with either raid expert or rst.
please be patientnow guys i see lots of people saying, whys this not banned, why is mine banned and not his, etc, please be patient, this will takes weeks, months, and will be ongoing, but it is no different than any other benchmark, other than the fact it has gone unchecked for so long.. if this happened in 3dmark11 it would be immediately checked and in the future pcmark05 will be the same
what is allowed and whats not?
-snip-
1. "Using software, performing hardware modifications or by human interaction altering the perceived speed of the benchmark program, tricking it to believe it ran faster and thus producing a better result. (adjusted 31st of August 2010)"
2. not modifying the benchmark itself
Yes d3d, video encoding via registry or powertoy and LOD go against these rules, BUT they have been used and accepted as officially OK, unless staff have specifically said something is allowed that goes against HWBOTs general rules then it is NOT allowed
if what you are doing in pcmark05 goes against one of those 2 rules, remove it now. staff will be contacting users with out of line submissions asking for proof on how to replicate them within the rules, if you are worried you cant provide that proof, remove your score now or it will be removed for you
use your brain guys, if you know something is not right, 2400MB/s virusscan on a weak AMD platform, then its probably not right, this is the ethical side of hwbot we can't enforce but hope is in place
Key areas we are targeting to start with
Some of the key areas we will be targeting, but not limited to:
Video encoding replacement, audio encoding
Irregularly large trans windows
Irregularly large memory latency
Massive browser scores
Anything out of line with file decoding
Crazy virusscans or hdd general
What is OK
I am not going to go through every tweak and say its ok or its not ok, thats what the general rules are for and they cover alot of the "super tweaks", but not all,
FD2 as a program is totally allowed, but using it to manipulate the window size or minimize windows is not allowed, similarly using a script to do this is not allowed as it breaks the "benchmark percieved speed rule"
-snip-
Raid XPERT
Still up in the air is raid xpert, we are discussing the outcome of this one right now, there and positives and negatives to both sides of the fence and we need to make sure everything is worked out properly
Massman said it was allowed, but now that we are aware that their is some bugging going on, we are reviewing it.
We don't want peoples expensive raid setups obsoleted by a bugged software,
Fairness
This is not a witch hunt, everyone will be given the oppurtunity to prove their tweak is fair in a confidential environment. If you tweak is legitimate, it won't be shared, it will be approved and you can continue using it privately if you so wish.
If you choose to not share the tweak with staff and your score looks out of line then it will be removed.
Lets hope we can clean the benchmark up once and for all, remember guys, please use your brain, we dont want to remove scores, so please do it yourself if you know they go against rules
-
-
i read it in a glaswegian accent and it makes me smile.
-
very solid run m8 on such an old card
-
haha i know what it scored, ty again btw, but what i mean is i'm sure this anomaly happens frequently. when someone hits a solid score but because they've used a random piece the boints shown in this snippet look 'meh'.
guessing that the point of showing recent members subs on the team page is to bump folks hard work i was suggesting unlinking it to hw points alone and maybe link the snip to the highest score on the submit, (be it hw, global or tpp). this then gives a better reflection of the submission in the snapshot.
-
shown in the pic below is my latest sub. it looks like a pretty ordinary run earning 2 hw points. the sub actually pulled 16 globals and was in the top 20 dual pc05 results.
would it be possible to change the boint display on team page have the highest result rather than just the uhp?
-
thanks guys. looking forward to improving the score over the next couple of weeks. still not used any irams for this run and i'm hoping the UP7 will bring something special to the party :celebration:
-
you sir are a gentleman :celebration:
-
thanks Turrican but i've tried recalcing as well as deleting and re-submitting with no joy. it's still listing it's rank in the 4 core cpu category (click on global points UGP and it goes to 4 core rank) link
normally it wouldn't be much of an issue but this one is top 20 global dual core
SniperOZ - 4x Radeon HD 7970 @ 1260/1750MHz - 8248.3 DX11 Marks Unigine Heaven - Xtreme Preset (DX11)
in Result Discussions
Posted
shit, wd fella. great score!