I can't wait to see the review. As I've told you and Kazoo on aim I think this will be a great addition to XS.Quote:
btw, the AMD Venice Preview from OPP should appear on the front page tonight! (i was too busy to post it yesterday, my bad )
Printable View
I can't wait to see the review. As I've told you and Kazoo on aim I think this will be a great addition to XS.Quote:
btw, the AMD Venice Preview from OPP should appear on the front page tonight! (i was too busy to post it yesterday, my bad )
Got the Venice up and running. Still trying to learn what I am doing. Thanks OPP for getting it to me by the weekend
C'mon Saaya, uploading has turned so complicated? :D
Ever feel like your being toyed with?
yes, yes i have, in fact, i kno ive been toyed with several times, so i pwnt the person toying with me. and are u gonna get us air results mebbe chunky?
OPP's Venice Preview is up on the Front page...
So what do you think, folks ???
LOL, that man is xtreme.Quote:
Originally Posted by saaya
G
its been on the front page already, sorry i didnt post a link here :DQuote:
Originally Posted by LowRun
http://www.xtremesystems.org/modules...owcontent&id=2
chunky, can you run some sse3 benches? :D
pifast with and without sse3 patch maybe?
Hey I dispose of $400 at the moment, do you think that would get me a 3800+ venice? because the newcastle is at $373 at newegg or only a 3500+ because of the initial price raise? :stick: im not much of an economist :hitself: :stick:
A Great Job , Very informative,and many THANKS to OPP"s .... :p:Quote:
Originally Posted by bachus_anonym
venice chips will cost the same as winnies atm if not less in a few weeks :)Quote:
Originally Posted by winston64
Saaya do you have a link to pifast? And the sse3 patch? I will run any benches you guys want. Just know that I am a terrible oc'er.
OK on patched super pi 16m I got 15m 34s
Hello !
AnyOne know when and where I can order me a 3500+ Venice ? I'm sure appreciated for that !
how-much will it cost an 3500+ Venice ?
It's much appreciated and a big thanks to all involved :up: Interesting too ;)Quote:
Originally Posted by bachus_anonym
sse3 or sse2 patched? ask kunaak, he knows all about those patches :DQuote:
Originally Posted by chunkylover77
ive only seen them on 2 shows in switzerland so far... they will cost a lot initially, but the prices should go down to winnie levels or lower soon :)Quote:
Originally Posted by AMDfan
and yeah, big thx to pc ice and oppainter again for those results! :toast:
I think it was prescott patched
i dont know if thats sse2 or sse3 :DQuote:
Originally Posted by chunkylover77
and i forgot to add! of course BIG thx to steven and chunky!!! :toast:
Could someone with a Venice run Prime95 benchmarks and post the results here (from results.txt or as screenie)? I'd like to find out, if AMD did something with SSE2. Together with the memory controller improvements this should be good for several % of increased performance.
hey, welcome to XtremeSystems :eleph:Quote:
Originally Posted by Dresdenboy
the venice isnt out yet, only 2 people here have a venice :D
the xbitlabs article is already out, they have a load of interesting results :)
amd didnt speed up sse2... i think thats what you meant right?
amd improved the imc and enhanced the data prefetch logic.
according to xbitlabs they also increased the write combining buffers from 2 to 4
(i thougt that was done in winchester already? :confused: )
and they only mentioned amds stretched silicon now for the first time, but amd has been using it for their 130nm cpus already afaik :confused:
Big day today ! release-time for the monster VENICE ! :YIPPIE: :banana: :toast: But no-one can tell me where I can find a 3500+Venice, :confused: I mean the web-sites, that may have it before anywhere else ? it would be nice if I can order it at soon as possible :slobber: Please, :worship: if some-one can give a link where I can order one of those 3500+Venice, I'll be very-appreciate it ! :eleph: :bounces:
Thanks :p:Quote:
Originally Posted by saaya
Well, AFAIK the write combining buffers were mentioned together with the other improvements by McGrath already in jan 04.Quote:
Originally Posted by saaya
What I'm looking for is if AMD changed something in the decoders so that they can decode SSE(2) loads in a way, that the 2x64bit access µOps can be executed by two different units in the FPU, not only one as it seems to be the case since first K8 (see
http://mersenneforum.org/showthread....ght=bottleneck and all the cache benches with 128 bit accesses).
And about strained silicon: DSL is a further improvement. They had simple strained silicon at AMD for 130nm.
Chunky, PM me when you get a chance. I'll try to answer all the questions you have about the cpu. Also, HEAT?? ;)Quote:
Originally Posted by chunkylover77
damn i see buying me this holy piece.
seems to do well under sub zero (hope will do ~-50°C) and this is rather interesting. my prometia has to be repaired and so my winchester on air suxx @ +50°C...will see what he does and then look if i get my nc 3500+ back and then this one - good plan -.-
Still NO one can give me the information, when and where to ordering those monsters VENICE ? :confused: :slobber: :(Quote:
Originally Posted by AMDfan
Unfortunately AMDfan, it appears that noone has them :(
Read this :)Quote:
Originally Posted by AMDfan
http://www.c627627.com/amd/athlon64
based on this, 3.0ghz and 3.2(san diego) by Q1 of 2006!!!!! u can't tell me the 4200+ San Diego won't do 3.0 stable on air(hopefully even the 4000+)
Wow, awsome results, and I was just getting settled in with my winnie, LOL... Out with the old, in with the new... Great job guys..
xbitlabs has a preview of the venice core as well. They show a break down of the +3800 and the +3500 for all the differnt cores; clawhammer, newcsl, winnys, and veince. I know someone was asking about a comarison of the differnt cores and there you have it...less than a 1% differnce. Like the thermal specs too. For a non-OCed chip, we are talking about some low wattge. In fact, thinking of building a server box on the venice, or sould I get a hyper threaded p4 (sacrelig I know, but every processor has its streghs and weeknesses).
Edit...wow never mind...the venice is kicking the newcsl's butt....go venice
ok guys. I have been reading up on the termal diode for the FX chip. There is a different programable value to control the boot of the processor. It seems that you can only program this at the factory. What that means for us is that the cold temp bug is not a bug at all, it is a direct result of factory programing. However, all is not lost. After looking at the power diagram, the diode has a negitave logic design. In other words, power when there is a problem and no power when when everything is fine. This means that a value of the termal temp waring boot diode has to be a vaule of 0, X, or Z to enable boot up. The thermal variunce has to be that of 100moh. What that means is if we can impeed the thermal probe, or just disconected it (remove the pin), that means no more cold start but.
I will also look into reprograming the Diode. This will be hard since that has to be a trade secret and AMD will not relece such sensitve information cause haxers could do VERY bad things if they had that information (like cause your computer never to boot). On the other had, all the information I need is in the pin plot they have availible online. However, it will take me actualy getting a venice and a FX chip (or a winny) to actualy test my results. That wont be happaning any time soon. I will locate the Pin shortly, so you you want to impeed the pin I find and find out, that would be fun, but i warn you, I am not at fault if you screw it up and take out a ground or power pin.
Table 29. Thermal Diode Specifications
Symbol Parameter Units Min Typ Max Notes
I Sourcing Currents µA 5 500 1
TOffset Temperature Offset °C 0 52 2, 3, 4, 5
1. The sourcing current should always be used in forward bias.
2. The temperature offset is used to normalize the thermal diode measurement to reflect case temperature at the worst
case conditions for a part.
3. This diode offset supports temperature sensors using two sourcing currents only. Single sourcing current
implementations are not supported by AMD.
4. The temperature offset is unique for each processor and is programmed at the factory. The diode offset value is found
in the Thermtrip Status Register described in the BIOS and Kernel Developer’s Guide for the AMD Athlon™ 64 and
AMD Opteron™ Processors, order# 26094.
5. TOffset should be subtracted from the temperature sensor reading. If the temperature sensor has an ideality factor
different from 1.008, a small correction to this offset is required. Contact your temperature sensor vendor to determine
if additional correction is required.
btw this is what I am looking into
3.6 Thermal Diode
The processor provides an on-die thermal diode with anode and cathode brought out to processor
pins. This diode can be read by an external temperature sensor to determine the processor’s
temperature. Refer to the AMD Athlon™ 64 939 Motherboard Design Guide, order# 30474, for
details on connecting the thermal diode.
3.7 THERMTRIP_L
The processor provides a hardware-enforced thermal protection mechanism. When the processor’s
die temperature exceeds a specified temperature, the processor is designed to stop its internal clocks
and assert the THERMTRIP_L output.
THERMTRIP_L assertion is only valid when PWROK is asserted and RESET_L is deasserted.
THERMTRIP_L assertion indicates the processor die temperature has exceeded normal operating
parameters. PWROK must be deasserted in response to a THERMTRIP_L assertion to enable proper
processor operation.
Once asserted THERMTRIP_L remains asserted until RESET_L is asserted.
If the processor’s die temperature still exceeds the thermal trip point when RESET_L is deasserted,
THERMTRIP_L will immediately be reasserted and the processor’s internal clocks stop.
so in other words. If silly MB manufatues use ABS(thermal probe) and see -90C as 90C then they send thermtrip_L signal so stop the clock (precived at a lock up). So we have to stop the MB from sending this input when a large negitive temp is sent.
Quote:
Originally Posted by Gsus_Freak
Which mean a bios mod could do it or what ?
nooo not again :shrug:Quote:
Originally Posted by Molester
why do üpeople keep using that site? i hope it finally dies off... those roadmaps are based on pure speculations and have been totally off over the last years...
they claimed amd would go for an fsb of 250mhz (or was it 266mhz?) in early 2004 or something...
please ignore that site! :D
Gsus_Freak there are two low temp bugs, the one related to the temp probe reading has ben fixed by some mobo makers already afaik (see kevins fx55 shows a temp of 3xx°C reading the temp probe) the OTHER low temp bug however is related to the imc afaik and the drive strength of the memory wich increases the lower the temp goes. :)
darn, thought i was onto something. O well, not many of us own a casscade anyway :) none the less a multi casscade setup (you guys got some cash flowing man). Not farmilier with the acronim IMC related to proccesor interface. I have heard of it refered to as integral multiplexor channel for IO and such (like a decade ago on some motrola chips), but I dont think that is what you are talking about is it? Anyway, nice score anyway, keep up the good work.
hahahahah i feel like a dope now. duhhhhh brain fart hahahah :)
the newest news:
ICE got new world record already:
http://valid.x86-secret.com/show_oc.php?id=10441
IS that my blue-black box cascade?:D
'Cold bug' my ass. Fantastic clocks Dave! :toast:
Lets see it put some of that power to the ground now, so to speak. :)
Then the secret is out!
For the record, apparently the 3800+ isn't bothered by the cold, while the 3500+ would boot at -40C but not -90C, doesn't seem to be related to the BIOS revision either :confused:
Also, the ~3.85GHz screenie is not benchable at present.
Whats 3.8ghz tho if we cant run 3d/2d apps at over 3.6ghz?
Come on Ice lets see some real world benches not that screen shot sh*t..
If you have it running enough to take that CPU-z and verifiy it you have enough time to run a 2d 1m or pifast benchmark..
enough of this Sreenshot crap!!!!
Xtreme My ass!! Act like the Name!! ;)
No offense as others will see my point here!!
screenshot or whatever, it is still rather impressive and hopefully a sign of what is to come on the 09 process.
Whoa, chill there - working with a three stage cascade and a brand new chip takes time . . . results will be posted as they come in.
just finish phone with Dave, he said he will get his effort in 4g..
this 3800+ loves cold and volt.:D
Yep, didn't think I should say anything (guys, please DON'T flood PC ICE with PM's, questions etc.) but saaya should be starting a new thread soon, the race to 4GHz with the 3800+ Venice is on! :party3:Quote:
Originally Posted by Onepagebook
Cool ss.
Just an FYI Gsus... the thermtrip_L signal is an output from the CPU, not an input. Other tidbit, it's function can be disabled.Quote:
Originally Posted by Gsus_Freak
Peace bro :D
ya your right. The output of thermtrip_L is a signal when 125C is reached. And when reset and thermtrip are active high signals then CPU goes into a halt state untill thermtrip is active low. Reset has to go to active low signal as well but gets reset by powc(or something like that, it checks Vdim and VDD and HTT links to make sure everything is ok).Quote:
Originally Posted by EMC2
that's great, 3.86ghz :D what vcore whas it done on?
Then the venice will be really good for xtreme cooling too at least if many chips is like this.
It would be cool if the venice would do be the first AMD to take 4ghz. :banana:
when we gonna see some benches on high speed?
insane man :DQuote:
Originally Posted by Onepagebook
respect to pc_ice :toast:
Still No new from VENICE ? ? ?
Guys, give PC Ice a break. He'll post results when he has them. Not everyone can spend their entire life in front of a computer.
it is crazy that the k8 is now getting as fast as net burt. With a shorter pipe and low latency ram, though, the k8 is a sure fire winner. I would like to see how fast that 3.8 can crank out number. Maybe it will be my new server box? Not gonna have 3.8 OP for awhile I bet. Only thing is you can't really multi-thread with a single venice. With the extra HTT interface disabled, it is imposible to dual chip them as well....sigh. Has anyone ever tryed to OC an OP? I wounder how good that would go. Basicly the same as an FX i would hope. Only thing I fear is that the Mem controler might not be as strong for OC'ing as the rest.
Anyway though, I envy you guys getting those chips so high. Never though I would see a .90nm k8 would do that, was thinking that the 45.nm stuff would. Anyway, hope to see some benchmarks soon enough. Hope your back is feeling better ice. Good luck hunting down the Low temp bug. It has been sugjested to me that by my Dad (he is an Electricl enginner) to have the chip running at a stable temp (in this case -40c) and gradualy lowering the temp. He said that on some of his old planes, they wouldn't be able to turn on the computer till it heated up. Electricity doesn't flow as well in cold, so maybe heating up the core and then droping the temps down will be a way to by pass this. Just a sugjestion, I don't even know if you have much control over your cascade. But if you do, you might want to try it out (though you might of already since you know around -90c it stop working).
People have overclocked Opterons before. Do a quick search and you'll see some results. The main problem is that most boards require registered/ECC RAM, and that stuff usually comes with higher latencies than regular stuff.
Actually, the colder it is the better electricity moves because there is less resistance. The reason the cold can sometimes have negative effects is because the electricity moves TOO well, and it can cause raise conditions and have signals get to places faster than they were designed too. :eek:Quote:
Originally Posted by Gsus_Freak
Remember this Guys you can tune a cascade just like stacking pelts.
So the base temp is low as hell but can't remove any real heat..
a 3.8ghz 90nm cpu at 1.75v =200w+ load easy sure, but Idle for a 2sec SShot maybe, maybe 100w 50% of the load wattage.
Also a -130c 3 stages oil will clog in seconds and drop evap temps as the load is put on the evap thus casuing the 3rd stage to fail..
So this isn't just a suicide SreenShot Its something thats just way out of stable lanes.
A 1m at 3.6ghz is 100x's as impressive..
Thats what makes Opp so great as when you see his max cpu clocks they are displayed in his result brower on the 3dmark orb.. Unreal Cpu clocks/Unreal Gpu clocks all done in sync as a Team always..
Now don't try to say Im flaming him as a person, as he's a great, great Cooling,overclocking, and tweaker we just really need to end these SreenShots that the max benchable speed is hundreds of MHZ less in real world app's..
There Is just no point..
Im happy he showed the world there is no Cold Bug as for that I praise him, but for the SShots "yes I will keep mentioning these" please put an end to it.. Its pointless. :)
:eek2: holy :banana::banana::banana::banana:!
EDIT: please dont post huge complex posts like that :D thx
-saaya
Don't be so quick to judge, you haven't seen anything yet ;)Quote:
Originally Posted by sandybeach
hey all, im new around here...i have just made the switch from intel to amd and purchased a new system....BUT WAS COMPLETELY UNAWARE OF THE 4 DIMM ISSUE AMD CPUS HAVE. What an idiot.
i PURCHASED A DFI SLI DR MOBO WITH 4X512 OCZPC4000 VX.
I havent purchased a cpu yet thank god cuz i didnt know what i wanted.
Did i just waste my money OR can the venice core save me? Will it oc with 4 dimms, or should i try and return the memory?
Where can i get a venice 3500 or 3800+? I live in canada.
Im not a noob but this post sure sounds like I am , haha.
You will be able to use 4x512, but only at 2tQuote:
Originally Posted by dsumanik
NCIX.com is a good place to pre-order Venice.Quote:
Originally Posted by dsumanik
Also, I thought it was a BIOS issue that 4x512MB can't be run at 1T with Venice yet. The CPU's mem controller is capable, but mobos aren't up to par. Wait for DFI to release a BIOS update before getting too upset.
:( No VENICE yet ? :( :slobber:
^ no. You'll know when it hits.
dsumanik ...
Getting 4x double banked dimms running beyond 200MHz is going to be tricky even for Venice, to ask if your VX will be compatable with a centain mother board and with an unavailable cpu is beyond prediction.
Well should i just return 1 gig of the memory then?
I guess i could always buy it later once we find out whats going on with the 4 dimm situation?
Its too bad i cant get my hands on a venice...id be willing to be a test mule and find out once and for all. I am sure there has to be some others out there who want this exact config im running....
Looks like i found the answer for myself:
From xbitlabs review:
"Athlon 64 processors based on Venice can work with four single-side DDR400 SDRAM modules without any limitations, and if there are double-side DDR400 SDRAM DIMMs installed, they can work at 400MHz with 2T timing."
:banana::banana::banana::banana:, looks like im going to be using 1 gig. Pc4000 VX is double sided im pretty sure.
My point is don't assume compatibility.... If you can't get 2 gigs of VX ram to work at your desired frequency/latency.... then sell the spare gig ;)
You just can't predict configuration compatibility with new hardware, wait for the early adopters to finish their beta testing.
dsumanki I have a great idea. Send me those sticks of ram and I can test if it works for ya. I'll send em back I promise.
Yeah, but the problem is that im a web designer, and my current intel system just died on me...Quote:
Originally Posted by bugeyes
i need a workstation up and running asap, every day is costing me money.so im in a hurry and unfortuntely at a hardware transition phase in the amd universe and i have money but no clue how to spend it. Im not worthy....
i used to have a
2.6@3.6 ghz 478
ic7 max3...which im sure is the culprit
gig corsair pro pc4000
i crappin my drawers cuz i had a bunch of recent work on a raid 0 array that i hope i can recover on the dfi.
i dont know if i should sell or hang on to the pc4000 I bet it overclocks nice at higher voltage...
on a side note, i have a "vtt fix" chip for an ic7 max3 i never got around to installing if anyone needs one out there. It offsets the improper vtt tracking above 2.8vdimm.
anyone who wants to have my ram ill trade for another evga 6800gt so im sli capable.
tee hee
oh ya hey guys no swearing on these forums i got in trouble already.....
the dutch guy a.p. versteeg has a San Diego that does 3300MHz at 1.45V,
cooled by a Vapochill LS.
Looks good to me :D
linkie
Nice... why is everyone hanging around 1.4v~ though, aren't they scaling with volts?Quote:
Originally Posted by Jack
Why don't some Pentium 4s like volts? It doesn't really matter, because they scale pretty well at low voltage, just like Venice seems to be doing.Quote:
Originally Posted by s0ck
nice overclocks do u think the fx-55 will still be good enough until the fx-57 comes out and will it be worth the upgrade too.
Ummm, didn't I just say this?? :)Quote:
Originally Posted by dsumanik
I would say tentatively, yes, assuming you have to have the top-of-the-line CPU and you have the money.Quote:
Originally Posted by Wayway
Right now we're all pulling these answers out of our behinds, because we haven't seen any FX-57 results.
No, you said 4 DIMMs means 2T, but that's not the case if the DIMMs are single-sided. :)Quote:
Originally Posted by s7e9h3n
This is my :banana: SanDiego 4000+ on almost stock voltage :D 3300MHz and still plenty of room to play, tomorrow I will try to go as high as possible with more Vcore and other memory.
http://home.wanadoo.nl/a.p.versteeg/3300lowVcorekl.JPG
omg impressive, how many cost? and where do you bought it?
cost? nothing> direct from AMD :DQuote:
Originally Posted by xavicl
hmmm, how come the CPU-Z recognizes San Diego and not Venice :stick:
By the way what cooling are u using there :confused:
waww.... and cooling? MachII? i wanna see results on air :(Quote:
Originally Posted by a.p.versteeg
It says San Diego because in has 1024kb of L2 cache which makes it a San Diego while a Venice only was 512kb of L2 cacheQuote:
Originally Posted by winston64
On cooling I'm guessing he is using his Mach 2 which is in his sig.
Cpuz recognizes VeniceQuote:
Originally Posted by winston64
why amd dont send me an fx57 ?????
for the same reason that they dont send you two FX-57s :slap:Quote:
Originally Posted by leviathan18
:rofl:
nope, a Vapochill LS ;)Quote:
Originally Posted by Cpt Twitchy
How does San Diego compare to Clawhammer (clock for clock/ipc wise)?
Newcastle -> Venice was a very big increase (where the biggest was the move to Winchester and 90nm), so following the same logic, shouldn't the new San Diego core show at least a few percent performance increase vs a Clawhammer (both clocked the same of course)?
Thoses Venices don't want to hit the streets or what ? :confused: :slobber:
still can't find them nowhere :(
AMD FAN,Quote:
Originally Posted by AMDfan
I can see your eagerness to lay your hand on Venice. Check this thread and ask the guy where is the local retail that have it in stock. http://xtremesystems.org/forums/show...d=1#post804819
Thank you for your kindness, I have asked the guy for his local store address or phone number, but I din't get any reponse yet :( I want it and I want it now :mad: :slobber: :slobber: :slobber: :banana: :banana:Quote:
Originally Posted by agenda2005
I want it too, I'm waiting for Venice to switch to s939 pci-e.
Relax and be careful. Don't go and send money to any phony place claiming to have venice. Be patient and I beleive the CPU will be everywhere by April 21st during AMD launch of Dual Core.Quote:
Originally Posted by AMDfan
Thanks ! I will keep it in mine ! :)Quote:
Originally Posted by agenda2005
should be faster, like 3% on average, and maybe up to 10% faster in sse3 apps at the same clockspeed :slobber:Quote:
Originally Posted by krille
dualcores are launched this month also?? lots of new cpus to choose from.... if only i had the $$ got to settle for a 3000+ or 3200+ venice.
edit: by the way in sor of new... been readign for a while and just did my first post... nice site, lots of info. :clap:
Oh god the Venice comes in and just blows away the winnie :D
I definitely want one. :slobber:
mmmm me 2
hi guys - so what is the final word on cold bug for venice? would an R507 modded prommy definiltey cause the cold bug? if so, is there anything I can do to stop temperatures going THAT low?