Tyan Toledo i3210W (S5211) - Workstation u serverskoj koži

[danij3l]

bubba?

treba li ocekivati josh review-ova serverske i workstation opreme? (pravo osvjezenje nakon tone Core2Duo overclock recenzija )

[Bubba]

Autor DaNiJ3L

bubba?

Jel' me netko trazio?

treba li ocekivati josh review-ova serverske i workstation opreme? (pravo osvjezenje nakon tone Core2Duo overclock recenzija )

Well, ovo je bio "prvi pogled", a ne review. Ajmo reci, "pilot clanak". Ako bude interesa i poklopi se jos nekoliko stvari, sto se mene tice, it will flow...

[Renesis]

a je bubba omastio brk

nego, jesu javili iz tyana što je sa tim grafičkim (ne rad, te rad na x1 ),te onaj pokrovčić na lan portu

[Bubba]

Autor Renesis

nego, jesu javili iz tyana što je sa tim grafičkim (ne rad, te rad na x1 ),

Stanoviti Mr. Andrew W. radi na reproduciranju moga problema. I'll keep you tuned...

te onaj pokrovčić na lan portu

Khm, tu je izgleda bila moja greska, al' nemoj nikome reci.

[tor]

Zaboravio saviti lim

[LAcy]

Autor Bubba

... it will flow...

Zašto mene ovo podsjeća na The spice must flow? (retoričko pitanje, Tor ne tražim odgovor iako ga znaš).

Autor tor

Zaboravio saviti lim

Buaaaaaaaa

[Bubba]

Autor LAcy

Zašto mene ovo podsjeća na The spice must flow?

Zato sto je i Duke Leto Atreides volio p0rn, tja, da spice, molim te...

Nego, jedno pitanje fur alles - cime bih najlakse (pod ovime mislim brojcano, tj. na neki konkretan 3D markoliki test) utvrdio ogranicenost PCI-E propusnosti. Svjestan sam da se radi o NV44 jezgri i o svega nesto manje od 5GB memorijske propusnosti, ali PCI-E 1x je 500MB aggregated bandwidtha a siguran sam da cak i GF6200 to moze iskoristiti.

Eh, naravno, totalno zaboravih - test bi morao moci biti u stanju izabrati na kojoj kartici se testovi izvrsavaju, dakako...

Ideje?

[tor]

Hm, nemam pojma ima li neki od programa tipa Everest tako nešto.

[Bubba]

Autor tor

Hm, nemam pojma ima li neki od programa tipa Everest tako nešto.

Mislim da ne - jasno ti je zasto pitam; ne pada mi na pamet skidati par gigabajta nekog 3D Marka koji ce davati 0.1FPS makar zalemio te NVSice direktno na procesor... Steta sto nije 8xxx serija u igri pa da iskoristim CUDA-u...

Evo zgodne stvarcice:

THE SERIOUS MAGIC TEXTURE DOWNLOAD BENCHMARK v1.0
http://www.seriousmagic.com

PURPOSE:
This benchmark was originally developed for internal testing at Serious Magic. We have released it publicly so that those who care about improving 3D graphics performance can see what we've observed. We believe it exposes a significant issue in PC graphics card performance. While today's graphics cards can render images very quickly, the software drivers are painfully slow at getting rendered output back over the AGP bus and into the PC where it could be saved and put to work by users. Current generation software drivers achieve only 1/100th of the theoretical download transfer speed that the hardware you've already paid for is capable of. It's remarkable that a graphics card with a video input and some video recorder software can record TV-quality images to the PC hard disk in real-time, yet the same card can't record it's own renderings at even 1/10th this speed. 

The problem isn't the hardware, it appears to be the software drivers. This is supported by the fact that the external video input to a VIVO-enabled graphics card can be moved over the AGP bus very quickly. The speed of transfering the 3D rendered results of the same card is very, very slow.  Yet it could be dramatically increased simply with revised software drivers. However, it appears that no manufacturer has yet made this aspect of driver peformance a priority. The first card manufacturer to address this performance issue would offer the following significant benefits to all the users of their hardware:

- Their graphics cards would become invaluable for rendering production output for TV, film and video. As things stand now, for typical video resolution images only about 5% of the time is spent on rendering. Why? Because today 95% of the time must be used just to transfer the rendered output back to the PC. This effectively nullifies the considerable advantages gained from the amazing speed of the graphics card hardware.

- Software could be written to allow users to actually record game output in real-time without much impact on game performance. After playing there would be a compressed movie of your game play saved on your HD. On a reasonably fast machine you could actually record your game output digitally via Firewire to your DV camcorder as you play or even compress and burn it to a Video CD or DVD at the same time you are actually playing. 

- Screen capture software that grabs motion images of user interfaces for the purposes of tutorials and training is a vital business application. Yet these useful recording tools are currently limited by the graphics card software driver to transfering only a few full-screen frames a second. It should be simple to record 30 frame per second output to your hard drive but this is unfortunately not possible due to the texture download issue. The first card manufacturer to revise their software drivers to address the issue will have significant appeal among businesses, training and graphics professionals.

- Despite the popularity of Internet streaming, it is not currently possible to stream real-time rendered output from graphics cards over the Internet. The network connections, PC processors and codecs are all fast enough today. Sadly, all of this horsepower is being held back by one remaining weak link; the texture download speed of today's graphics card drivers.

Ova 1x daje ~150MB/s dok ova na 8x daje ~1050MB/s...

Svejedno bih volio provrtiti kakav maleni 3D testic cisto da jos i to potvrdimo.

[Jerry]

Analiza PCI Express skaliranja (1x-16x)

novija verzija (8800GT)
http://www.tomshardware.com/reviews/...ysis,1572.html

starija verzija (X600, X800, 6800)
http://www.tomshardware.com/reviews/sli-coming,927.html

P.S.
Zasto bi s grafom testirao propusnost (kada toliko vec mrzis 3D)?
Ima i drugih (vjerujem tebi puno drazih) stvarcica koje idu u PCI-E.

[Bubba]

Autor Jerry

Analiza PCI Express skaliranja (1x-16x)

novija verzija (8800GT)
http://www.tomshardware.com/reviews/...ysis,1572.html

starija verzija (X600, X800, 6800)
http://www.tomshardware.com/reviews/sli-coming,927.html

I sto bih ja sad trebao s time? Skidati TriDeMark Ventiđ i vrtiti ga na NV44 chipu?

P.S.
Zasto bi s grafom testirao propusnost (kada toliko vec mrzis 3D)?

O Jerrotije, opet nisi pio tablete?

Ima i drugih (vjerujem tebi puno drazih) stvarcica koje idu u PCI-E.

Ne znam kakve su tvoje sklonosti prema hardveru, no meni su sve "stvarcice" vezane uz IT jednako (ne)drage. U ostalom, sto da stavim u PCI-E kad su oba puna?

Da zatomim tvoju znatizelju - 250(*2) MB (== PCI-E 1x) nije dovoljno za vuci dva monitora visoke razlucivosti + nekakav stream ili tesku grafiku na njima. Sve da i jest, platio sam 2x PCI-E 8x pa ih zelim makar ih nikad ne iskoristio.

[Jerry]

Autor Bubba

I sto bih ja sad trebao s time? Skidati TriDeMark Ventiđ i vrtiti ga na NV44 chipu?

Ne moras ti nista skidati i vrtiti... sto se mene tice. To je bio samo prijedlog. Valjda ti netko moze to posuditi na CeDeJu.

O Jerrotije, opet nisi pio tablete?

Ma neee...
Danas ima puno boljih stvari!
I mislio sam na 3DM ali omaklo se.

Ne znam kakve su tvoje sklonosti prema hardveru, no meni su sve "stvarcice" vezane uz IT jednako (ne)drage.

Ako ti tako kazes...

U ostalom, sto da stavim u PCI-E kad su oba puna?

I komp ti nece raditi ako privremeno oslobodis jedan slot?

BTW, sve u svemu si me potpuno krivo razumio za ovo po pitanju hardwarea ali nema veze.

[Bubba]

Autor Jerry

Ne moras ti nista skidati i vrtiti... sto se mene tice. To je bio samo prijedlog. Valjda ti netko moze to posuditi na CeDeJu.

Gle, sve da mi i zavrtim nesto ovakvo - NV44 + 64bit memorijska sabirnica, helou? Nije bas neka sreca - a pouzdano znam da 3D Mark 2001SE (recimo) *ne zna* raditi s dvije kartice (tj. daje izbor, ali benchmark vrti samo na primarnom monitoru, da ga yebesh).

Nasao sam nekakav... "PerformanceMark", i uglavnom, vidljiv je pad performansi, ali taj program mi je sve samo ne pouzdan...

Ma neee...
Danas ima puno boljih stvari!
I mislio sam na 3DM ali omaklo se.

Onda my bad.

I komp ti nece raditi ako privremeno oslobodis jedan slot?

Hoce, ali:

a) sto da uguram unutra? Nemam viska PCI-E kartice na kojima bih mogao izguziti veliki bandwidth.
b) sve da i imam, mjerenje bandwidtha na negrafickim karticama je uzasno komplicirano.

BTW, sve u svemu si me potpuno krivo razumio za ovo po pitanju hardwarea ali nema veze.

Posipam se ostacima AMD-ATI biltena.

[Bubba]

U nadi je spas, kazu... No, za nadu i svasta drugog kazu...

Hello,

I have sent your results to the BIOS team. I don′t know how long it will take for them to produce a new BIOS, I will let you know when they do.

Thanks,
Andrew

[Bubba]

Hello Bruno,

There has been an issue found with the chipset this motherboard, and all motherboards using the 3210/3000 chipset. The PCI-e slots connected to the chipset will only run non graphics solutions at full speed. When you use a video card on the chipset it will clock down to x1. Intel did this so that you are forced to buy a workstation chipset for graphics cards. Since this is a server chipset and the motherboard has onboard video, the design was not for video applications. We have now been instructed to make the BIOS show both slots at x1 speed.

Andrew

Evo i moj odgovor, zivo me zanima sto ce mi reci na ovo:

Greetings,

I can only say that this was expectable, knowing Intel and their polices. However, although Intel suggests it does not support PCI Express graphics (qoute from Intel's i3210 datasheet, page 22; "The 3200/3210 MCHs do not support PCI Express graphics"), it still posses PCI Express graphics signaling as well as the manual clearly claims that there is a "PCI Express graphics attach" present on the board, that is obviously used for that very same graphics that they are attempting to diminish. No degradation in bandwidth or anything similar is mentioned in any form of whatsoever. Is there anything intrinsic in PCI Express graphics that would technically prevent them from giving full bandwidth? According to "PCI Express Base Specification, Revision 1.1" that they provide as a reference point of their PCI Express implementation - not really. In other words - quite dirty game from Intel...

So, to get it straight - as much as I actually meant to help and reported a "bug" (more like a feature, it seems, after all), it will result that the next revision of BIOS will actually make things even worse - lowering both PCI Express lines to 1x, opposing to the status quo that had (at least) one PCI Express slot fully functional? So, will this be applied retrogradely to all previous BIOS versions? Is this even going to be mentioned as a new "feature" of next BIOS version? Naturally, there was absolutely no harm done by you (Tyan) however Intel has forced vendors to break not so splendid news to their buyers who have already bought Intel based products without knowing this fact. It is, though, quite special situation (two discrete graphic cards on, in terms of market nomenclature, "server board), but still...

In any case, thank you once again for your swift and straightforward reactions.

Regards,

Bruno

[Baja 001]

Autor Bubba

Hello Bruno,

There has been an issue found with the chipset this motherboard, and all motherboards using the 3210/3000 chipset. The PCI-e slots connected to the chipset will only run non graphics solutions at full speed. When you use a video card on the chipset it will clock down to x1. Intel did this so that you are forced to buy a workstation chipset for graphics cards. Since this is a server chipset and the motherboard has onboard video, the design was not for video applications. We have now been instructed to make the BIOS show both slots at x1 speed.

Andrew

Izgleda da te je Bubba dobro namagarčilo, treba si ostat na Bad Axe!

[Roberto]

Autor Baja 001

Izgleda da te je Bubba dobro namagarčilo, treba si ostat na Bad Axe!

Sumnjam da su ga baš namagarčili, stvar je skoro pa irelevantna - za grafe gotovo da i nema razlike u performansama PCIE x1 vs PCIE x16.

Ipak mi je nerazumljivo zašto to nisu mogli napraviti kak spada.

[Bubba]

Autor Roberto

Sumnjam, stvar je skoro pa irelevantna za grafe gotovo da i nema razlike u performansama PCIE x1 vs PCIE x16.

I te kako ima. Pa covjece, o cemu ti pricas? PCI-E 1x je ~250MB, sto je ekvivalent ~66MHZ 32 bit PCI. To je *premalo* cak i za bilo kakav ozbiljniji 2D (dva monitora u visokoj rezoluciji + overlayi), let alone za nekakav potencijalni 3D. Uostalom, razlike u 3D benchmarcima, cak i sa ovim jadom od underclockanog NV44 chipa i polumemorijom razlika izmedju PCI-E 8x i PCI-E 1x je vise nego mjerljiva.

[danij3l]

nije problem u "nisu mogli"

nisu htjeli. to je za njaih serverska ploca i kao takva nema potrbe za grafickom koja treba vise od 1x pcie bw. (kao sto pise gore sa workstation plocama taj "problem" ne postoji)

[Roberto]

Autor Bubba

I te kako ima. Pa covjece, o cemu ti pricas? PCI-E 1x je ~250MB, sto je ekvivalent ~66MHZ 32 bit PCI. To je *premalo* cak i za bilo kakav ozbiljniji 2D (dva monitora u visokoj rezoluciji + overlayi), let alone za nekakav potencijalni 3D. Uostalom, razlike u 3D benchmarcima, cak i sa ovim jadom od underclockanog NV44 chipa i polumemorijom razlika izmedju PCI-E 8x i PCI-E 1x je vise nego mjerljiva.

Napisao sam to tako jer mi i s tim problemom ta ploča izgleda kao jedna od boljih u kategoriji server za workstation.


Hm, za 2D ne znam koji software koristiš, ja osobno takve razlike nisam primjetio na svojoj kofiguraciji kad sam isprobavao stvar. Jest stvar je bila mjerljiva s 3DM, i to je uglavnom to. Ja koristim jedan monitor, ne znam kako bi to bilo s dva, da li b irazlika bila veća.

Nisma nikad pridaovao puno značenja brzini PCIE sabirnice, pogotovo kod ploča koje su primarno namijenjene serverima.

[Baja 001]

Autor DaNiJ3L

nije problem u "nisu mogli"

nisu htjeli. to je za njaih serverska ploca i kao takva nema potrbe za grafickom koja treba vise od 1x pcie bw. (kao sto pise gore sa workstation plocama taj "problem" ne postoji)

Je oni su tako ali striktno tako zamislili MBO, nije im palo da će je netko uparit i sa jačom grafičkom kako bi napravio neki Workstation i za nešto drugo gdje samo CPU/RAM ne igra ulogu. Svakako šteta, potez je

To je isto kao da kupiš terenca, a da nemože vozit po cesti

[tutix]

Čitam i ne vjerujem. Prljava igra, ništa drugo. Pretpostavljam samo da je ekvivalentna workstation ploča daleko skuplja. Intel, Intel...

[Bubba]

Autor Roberto

Napisao sam to tako jer mi i s tim problemom ta ploča izgleda kao jedna od boljih u kategoriji server za workstation.

Za single socket ploce - vjerojatno si u pravu. Ipak, PCI-E 1x nije dovoljan za bilo kakvu suvisliju grafiku.

Hm, za 2D ne znam koji software koristiš, ja osobno takve razlike nisam primjetio na svojoj kofiguraciji kad sam isprobavao stvar. Jest stvar je bila mjerljiva s 3DM, i to je uglavnom to. Ja koristim jedan monitor, ne znam kako bi to bilo s dva, da li b irazlika bila veća.

Dovoljno je da stavis neki tesko nabudzeni PDF na dva monitora, 1600x1200, 32bit i *osjetis* (dakle, ne "izmjeris") razliku.

Nisma nikad pridaovao puno značenja brzini PCIE sabirnice, pogotovo kod ploča koje su primarno namijenjene serverima.

Cuj, Tyan je mogao uredno staviti 2 puta 8x fizicki PCI-E slot i dovidjenja. Nisu imali *nikakvog* razloga za staviti 2 puta 16x fizicki slot osim ako nisu ocekivali da bi netko mogao utrpati graficku karticu unutra, jer AFAIK, "najsire" negraficke kartice za PCI-E su jos uvijek 8x i vjerojatno ce i ostati takve.

A kao sto sam i napisao, Intel nigdje izricito ne spominje sto se dogadja kad i ako stavis graficku karticu na njihov chipset.

Autor tutix

Čitam i ne vjerujem. Prljava igra, ništa drugo. Pretpostavljam samo da je ekvivalentna workstation ploča daleko skuplja. Intel, Intel...

Mislis da je AMD bolji? Pogledaj razliku u cijeni izmedju 2xxx i 8xxx serija Opterona, samo za pocetak. Da ne spominjem kako su dual/quad socket ploce prakticki beskorisne samo s jednim procesorom, jer pola funkcija na ploci tada ne radi i slicno.

Ovo je slucaj kada je Tyan (a valjda i ostali) stvarno stavljeni u govna. Bas me zanima hoce li mi lik odgovoriti na pitanja...

[Roberto]

Autor Bubba

Nisu imali *nikakvog* razloga za staviti 2 puta 16x fizicki slot osim ako nisu ocekivali da bi netko mogao utrpati graficku karticu unutra, jer AFAIK, "najsire" negraficke kartice za PCI-E su jos uvijek 8x i vjerojatno ce i ostati takve.

Da, kad tako pogledaš na stvar, ovdje si potpuno upravu. Besmislica.

Baš me zanima kaj će reći na kraju, ako će uopće i odgovoriti.

Što se ostalih tiče, u ovom poslu kao i mnogim drugima, nema svetaca. Svi će oni prodati dušu za bolju prodaju i pritom poželjet da onome drugome ''krepa krava''. Primjera ima kolko hoćeš.

[Bubba]

Hello Bruno,

I was really surprised to find out that Intel did that with their chipset, I argued fiercely for many days to prove that it is the case. It is a shame that Intel would practice such shady tactics.

The BIOS′ will not be retroactively changed to my knowledge, would require too much work. I also doubt they will note the ″fix″ in the patch notes. I do know they are working to amend the website to display the information correctly. It also has been proving pretty hard to ″fix″ when you have two video cards you can trick the system into running one at x8.

Andrew

Mamicu in yebem. Heh, neka, neka, sad nek se iskrve dok ne natjeraju tu sugu od Intelovog chipseta da skrese OBJE graficke na 1x, ehehehe...

Eto, EOD. I lijepo je za znati...

[Bubba]

http://tyan.com/support_download_bio...?model=S.S5211

Problem Fix Description:  
 - Fixed an issue where some 3Ware Raid controller cards wouldn’t be able to load their Option ROM during POST
 - Fixed an issue where you can use a DOS display screen to access a M3296 functions
 - Fixed an issue where disabling the onboard VGA would cause an 0260 error during POST
 - Updated embedded VGA Option ROM to v1.09.10
 - Updated the  PECI CPU display information
 - Fixed an issue where watchdog timer functions would fail during a Power ON state
 - Fixed an issue where watchdog timer would cause the system to NO POST (98h) when enabled.
 - Updated MRC to v1.41

Iju iju ijuju, 'ko bi reko da ne bude pisalo kako su pokusalu ukebati PCI-E bandwidth za graficke, hehehe.

Doduse, dodali su pod opis ploce kako nema podrsku za puni PCI-E bandwidth u slucaju koristenja graficke kartice; ako nista, gadovima ne gine reprint manuala za nove ploce.

[Bubba]

E jesu specijalci ovi u Tajanu...

Gledam ja neki dan memory compatibility listu, kad ono, 4GB moduli se spominju kao testirani. Kako po specifikacijama chipseta i ploce pise da je maksimalna podrzana kolicina RAM-a 8GB, uz lagani slijeg ramenima zapitah se zasto bi, dovraga, isli testirati 4GB module kad ploca ionako podrzava "samo" 8GB.

I ne bi ja lijen...

=== REQUEST SERVICE Bruno Banelli: 09/01/2008 23:38
Greetings,

according to the list on this page:

http://www.tyan.com/support_download...?model=S.S5211

Tyan Toledo i3210W appears to support 4GB memory modules. Is that information correct and thus, can two 4GB modules operate in motherboard, giving aggregated value of 8GB memory (maximally supported by the chipset). I trust that adding any additional memory modules would prove to be useless?

I am inquiring since it seems odd that motherboard with 4 available slots and maximum of 8GB memory capacity would support such large modules. In addition to that, Intel′s datasheet for Intel 3210 chipset has no explicit statements about the support for such high density modules.

Thank you in advance for your assistance,

Bruno Banelli

Kad li, unatoc mojem low priority pitanju, stize odgovor:

=== UPDATE Rodger Dusatko: 09/02/2008 11:39
Hello Bruno,

4Gb memory modules were not so common at the time when this board was first developed. For this reason, the 8 Gb (4x 2Gb) was tested then and found successful. Since we did no testing at that time for 4Gb modules, we wrote in the docu that only 8Gb were supported.

Now that the 4Gb modules from Kingston have been successfully tested (KVR667D2E5/4G with Elpida E2108ABSE-6E-E chips), with these modules 16Gb is possible. All of our tests require that we fill all DIMMs with the memory. In this case, 16Gb of this Kingston memory were verified as compatible and functioned properly.

Rodger

U svakom slucaju, jos jedan prilicno veliki bonus za ovu plocu, jer mislim da je to jedna od rijetkih Socket775 ploca koja podrzava 16GB memorije. Vidim da se Kingstonovi 4GB "vulgaris" (non-ECC, dakle) moduli prodaju za tricavih 600njak kn, pa cu probati iskamciti od nekuda 4 komada, makar za probu, iako se bojim da, ako budu radili kako spada, ce ostati kod mene...

[Renesis]

bogme, support im je na visini