Printable View
AMD is announcing a revision change for its:
Desktop AMD Athlon(tm) 64 processors from revision D0 to E3 for model numbers 3500+, 3200+ and 3000+ in socket 939.
http://www.amd.com/us-en/assets/cont...hange_1_UK.pdf
Desktop AMD Athlon(tm) 64 processor 3800+ from revision CG to E3.
http://www.amd.com/us-en/assets/cont...hange_2_UK.pdf
Desktop AMD Athlon(tm) 64 processor 4000+ from revision CG to E4.
http://www.amd.com/us-en/assets/cont...hange_3_UK.pdf
So it begins...
what does?Quote:
Originally Posted by Ackbar
...IT...Quote:
Originally Posted by 3rd-Dimension
That's kinda funny... Just ordered a 3000+ 939 on sunday.
Hopefully these E3's will be worth upgrading to.
I think he's talking about the hysteria/hype the new revision brings...Quote:
Originally Posted by 3rd-Dimension
There will be a 1,000 new threads about these in the next few months.
4th of april...let the countdown begin :toast:
Yeah, and I know 100+ people who are dreaming that they will clock up up in the sky ...
And what are advantages of new revision? SSE3, ...?
There are rumors a plenty, the only thing we know for sure is SSE3. Some people say sSOI... who knows. I don't see any particular reason it would be able to reach ridiculous overclocks unless it was sSOI and even then who knows.Quote:
Originally Posted by Moonman
hmm, appril 4th - the question is should i wait or not. i am building a new system, and have yet to buy a cpu. how reliable is the april 4th ship date, and how soon can we get our hands on them after they ship?
What are the differences between rev E3 and E4?
well using a bit of common sence i would say E3 = 512kb cache and E4 = 1mb cache :)
other speculation includes an upgraded/improved memory controller...
i guess we'll see how valid this is soon enough.
I wonder if the lower end will not clock as high since they are shrinking to 90nm on the higher end chips. Speed binning might mess it all up! before there were only 3 chips to compete with. 3000+, 3200+ and 3500+.
Now they are adding more chips to the mix and it is going to get more competitive. Do you get what I'm trying to say?
Where did you get this magical number?Quote:
Originally Posted by Charloz24
Figures. Just got my 3200 last week. Bah.
In the link above? :PQuote:
Originally Posted by RAINFIRE
Well let's hope the wait(s) are worth it.
You could be on to something.Quote:
Originally Posted by trans am
I imagine they speed bin by spec. For example all the processors that are stable at 2200mhz stock volts get labelled 3500+, How many winchesters where stable at the next speed grade using stock volts? A bunch I'm sure...
Well now with the new venice core's they will just slap a 3800+ label on the best one's.
I cant wait to test a new 3200+
You have to think that more 3000/3200 +'s get sold tho, so even higher ones will have to be downgraded to keep up with the demand, so everyone flood them when they come out :)
I wonder what percentage of each wafer meets each speed grade, and what percentage needs to be downgraded to meet demand?
I am sure this is a closely guarded secret!
Speed binning isn't quite as bad as you are all making it out to be. In my advanced CMOS course our professor has discussed it in length as well as shown us a few other sources. Apparently, the vast majority of all chips on the same wafer perform the same, if anything, the ones on the outer edge of the silicon wafer are slightly less capable, but not by much and rarely do they have to downgrade the part.
Just hope that you aren't the unlucky sap who gets one of those few that are downgraded. With my luck that's what I'll get.
another thing to consider.. i think amd also bins thermally.
if a chip is over the TDP at a speed it's stable at, it'll be bumped down to meet the TDP spec.
and then you have to consider that the mobile chips will get the best thermal bins. and the fx should get the highest speed bins that make the 89w limit.
just my theory though :sofa:
edit: orthoganal. i'd believe that, but that doesn't mean that all wafers are the same :toast:
That's quite right, I didn't add that little caveat to my explanation, but the quality of the silicon does make a difference. My comments were directed more at the quality of all the chips from a particular wafer....Quote:
Originally Posted by ozzimark
btw, that reminds me, I have a test in that class tomorrow, I better get studying.
Do AMD not take all there winchesters from the same wafers? Or are there differrent wafers for each speed grade? Chips are not Speed binned until asembled in Malaysia.
Hmm, I agree that it is the silicon which is the mayor difference, let's hope the new supplier will cease his streak of bad 05 chips and bring out something SWEET :)
Lets just hope strained silicon is all it's cracked up to be.
so if i order april 5th from new egg ill get Rev E3? Might as well wait a few more weeks for the super hype chips
so is this venice then :confused:
Perkam
It's most likely San Deigo since it's 1mb lv2 cache, but it's basically the same thing. Venice is the 512k lv2 cache version.Quote:
Originally Posted by perkam
iirc..
troy = 1mb, 940
san diego = 1mb, 939
venice = 512kb, 939
newark = 1mb, 754
lancaster = 512kb, 754
i previously thought that newark would be the 1mb DTR chip while lancaster would be the 512kb low voltage mobile.. but it seems that it's not actually working out this way, so i believe those are what the codenames represent :stick:
What perf increase are we really going to see with the Venice...I mean the Winnie's are stretched out as far as they can go (up to 3ghz) as it is, so they cant OC better. So then does this mean higher 3d marks at same clock ???
Perkam
remember seeing that thread with the 1.6ghz turion doing 51.9s in superpi 1m?
my clawhammer gets 52s (normal non-natched) at the same settings... so i'm not expecting a noticeable change at the same speed.. just the top end to go up a lot :banana:
They better overclock well. I'll be quite disappointed if they don't. :(
Speed binning might be a bit of a concern, but I'm thinking it'll be more like 6800GT clocks vs. 6800 Ultras (not THAT big a difference).
Either way, with all the new optimizations and doo-dads they're tossing into this new revision it's probably not worth it to buy a Winchester right now.
all i want is 2.7 after that is just icing on the cake. Give us a mem controller that will handle anything we throw at it. PLEASE
hopefully this switch to .09 to most of the processors, doesnt mean we get a whole lot of CPU's that can't be used for sub zero stuff.
the E revision chips are supposed to remove the 4 sticks = 2T and allow full banks/dual sided ram to run at 400MHz, rather than locking it down to 333, as well as providing more compatibility, SSE3, 90nm (for those that don't already have it), and hopefully price drops (more CPUs per wafer = lower die cost... hopefully reflected in pricing :) )
Venice is coming out on 5th San Diego 4000+ on 12th. As for the comment about silicon and marginally worse on the outer edge I heard the opposite and that most defects come from the centre. The comment about speed binning being done in Malaysia I can confirm.
So I get to ditch my poor OCing FX-53 for a *possibly* uber OCing San Diego next month? Sweet. Give me one that I can run at 12x250 and I will be happier than a pig in the mud.
Is Troy more susceptible to Trojans? :DQuote:
Originally Posted by ozzimark
ATM I could care less. I'm extremely happy with my 754 3000+ setup and it'll last me at least another year or two. I can't wait *rolls-eyes* for all these new chip threads that'll be popping up all over the place.
Woohoo :) Now I await CEBIT
About the Troy, that is the 2 way (what we see now in the 252) the 1 way is the Venice and the 8 way is the Athens.
I suppose you mean "couldn't care less", right? ;)Quote:
Originally Posted by scrible88
Anyway, this is good news! With the reports of the 4200+ doing stock speed (2.6GHz) at 1.25V, I suspect that we're going to se a few AMD O/C records. As long as they like subzero temps, that is. Well, I'm probably still going to upgrade to a really cheap Sempron 90nm now and go A64 next year. Gives me something to play with in the meantime! :)
I just returned my 3000+ Winchester.
Do you think i should wait another month and get a Venice? Will prizes rise?
Prices tend to drop each quarterQuote:
Originally Posted by DildePer
So what you're saying is that the Venice core should be cheaper than the Winchester?
so it seems. I'm waiting. I just hope they have their :banana::banana::banana::banana: together on the first runs.
I very much doubt that the prices of venice's will be lower to start, remember the winchester release, everything went higher and higher then dropped. And this is "major" compared to just moving to 90nm..........or is it? is sse3 actually used by that many prog.s atm, i was under the impression it wasnt.
I think AMD intends for this release to be semi-major. SSE3 offers some optimizations but nothing tremendous. Winchesters are aleady 90nm, so they're basically just adding SSE3 and (speculation) improving the memory controller/strained silicon. So it's possible they're only adding SSE3 at the moment and absolutely nothing else since that's all we're sure of. If that's the case, I don't see that this release would be a huge deal for AMD.Quote:
Originally Posted by Vincentvega18
damn it just bought CBBHD 0450 :(
should i sell it ???
Quote:
Originally Posted by TEDY
If it's a good clocker, no.
Venice cores will be out for some time, your not in a hurry :)
That's a pretty decent overclock for such low voltage! I'd say keep it, but if you're looking to upgrade you could always PM me... ;)Quote:
Originally Posted by TEDY
Quote:
Originally Posted by TEDY
I would say no. You have a great OC in your sig, and with the low voltage, room to go higher. Since it is only speculation anyways that the venice will OC better, why chance it?
Its Athens, Troy, Venus for 1 way, 2 way, and 8 way.Quote:
Originally Posted by OC Detective
Exactly! Although I am glad new ones are coming out. It just means cpu's, etc will be more advanced by my next upgrade!Quote:
Originally Posted by Mikael
w00t
Quote:
Originally Posted by Moonman
the hype has already started, theres one dude already talking about getting 400x12=4.8GHz :stick: everyones making fun of him now, im telling him those arent p4s and to expect 3GHz to be a good goal with air, maybe 4GHz with cascade or dry ice :stick:
"I don't see any particular reason it would be able to reach ridiculous overclocks unless it was sSOI and even then who knows."
some say all will be strained silcon, others say just the diegos. One guy thinks amd will put OVP and maybe even multilock their 512k venice chips to limit their overclocking while leaving the diego open for xtreme overclocks cause amd makes so much more $ on diego, it to amd's advantage that people do NOT buy venice instead of diego to overclock so I guess those who dont care to oc much go venice, those that want good oc go diego. I am hearing that diego may do 200+ higher on average than venice then you also got double the cache to throw in as well.....
"Speed binning might mess it all up! before there were only 3 chips to compete with. 3000+, 3200+ and 3500+."
amd may bin the best silcon for diego and make sure the venice, especially the lower ones dont get as good bins. Thus 3000+ and 3200+ is likley to do worse than 3500+ and up, this is already true for many of the most recent winchesters. Thing is I am not too keen on spending almost $300 for a 3500+ venice or twice that for a 4000+ diego, I mean my $160 winny does 2.6GHz
"Figures. Just got my 3200 last week. Bah."
theres always ebay ;)
"You have to think that more 3000/3200 +'s get sold tho, so even higher ones will have to be downgraded to keep up with the demand, so everyone flood them when they come out"
amd could do even more testing with higher clocks and undervolting, maybe if a 3500+ needs 1.3v to do that and another needs 1.2v amd will underclock the 1.3v one and sell as a 3000+ so youll still not get as good an overclock as a "real" 3500+ yes your 3000+ will still hit 3500+ speeds but the 3500+ may do more :D
"Apparently, the vast majority of all chips on the same wafer perform the same, if anything, the ones on the outer edge of the silicon wafer are slightly less capable"
perhaps but what about different wafers? amd could use the best wafers for the highest binned venice and for all diegos and they can see if a wafer is not giving them high clocks, they will bin all chips on that "weak" wafer as 3000+ and 3200+
"They better overclock well. I'll be quite disappointed if they don't."
they probably will if you get a diego or a 3800+ venice, amd may speedbin the lower ones. it all boils down, would you pay $200 for a 3200+ venice to get 270x10 or $400+ for a 3800+ Venice to break 3GHz? Want 3GHz and 1mb cache? Thatll set you back $600+ for 4200+ diego or near a grand for the exclusive fx57! :slobber: as for me, I am getting either the 3000+ or 3200+ Venice depending how they clock. I wont need 10x multi if they dont clock high enough, ill just run 300+ fsb at 9x 1:1 thanks to its improved ondie controller.
"all i want is 2.7 after that is just icing on the cake. Give us a mem controller that will handle anything we throw at it. PLEASE"
ondie controller is a big factor for me, my winny cant get past 262 with my 2x512 patriot, id love to run 300x9 with that venice. I do want at least 2.8GHz though seeing my winny does 2.6GHz, seems kinda silly to sidegrade just for the ondie controller and besides I already do 289fsb with 2x256mb ram which is much easier on the ondie controller.
"So I get to ditch my poor OCing FX-53 for a *possibly* uber OCing San Diego next month? Sweet."
2.5GHz is indeed poor for a fx53, id ebay it now before diego comes and lowers its value.
"I'm extremely happy with my 754 3000+ setup and it'll last me at least another year or two."
I used to also own a 3200+ s754 at 2.58GHz but it cant hold a candle to my winny
"With the reports of the 4200+ doing stock speed (2.6GHz) at 1.25V"
this doesnt tell me much, undervolting is not = to overclocking.
"They better overclock well. I'll be quite disappointed if they don't."
they probably will if you get a diego or a 3800+ venice, amd may speedbin the lower ones. it all boils down, would you pay $200 for a 3200+ venice to get 270x10 or $400+ for a 3800+ Venice to break 3GHz? Want 3GHz and 1mb cache? Thatll set you back $600+ for 4200+ diego or near a grand for the exclusive fx57! as for me, I am getting either the 3000+ or 3200+ Venice depending how they clock. I wont need 10x multi if they dont clock high enough, ill just run 300+ fsb at 9x 1:1 thanks to its improved ondie controller.
I think if 3200+s did an average of 270x10 that would be a good improvement. Winiies do it, but most dont.
Magnj
>>Want 3GHz and 1mb cache?
There's always the 4000+ too.
anyone know what the prices will be roughly?
Same for same rating. Winchester should cost about the same as Venice before running out of stock.
They posted new prices today. FX-55's look like most significant drop.
AMD Pricing, March 10th
Not all of us have unlimited money to spend on our systems. Especially as a high school student. It does everything I need it to and does everything I need it to well. I don't give a crap if it "can't hold a candle to your winny"Quote:
Originally Posted by Geforce4ti4200
Its just not a worthwile upgrade for me. I don't have money to burn.
Quote:
Originally Posted by EMC2
AMD Athlon™ 64 FX-55 $827
4000+ $643
3500+ $272
3200+ $194
3000+ $149
Still not cheap. If that 3000+ is a venice, thats a sweet price. The 3200+ is alot more so Venice better clock high for it to be worth the 10x multi.
Big price difference going to 4000+ from 3500+ for 200MHz more and 1mb cache.
"Not all of us have unlimited money to spend on our systems. Especially as a high school student."
That I understand, but I had the $50 to go from s754 to s939 after all was said and done ebaying my s754 stuff. I am ebaying my winny so Venice will be a cheap upgrade too.
"Not all of us have unlimited money to spend on our systems. Especially as a high school student."
Amen... 7.50$ an hour aint cuttin it
try 3.5$/h i'm getting here...
nah its just inhabited by them ;)Quote:
Originally Posted by Ackbar
im really excieted about this, now i will feel like i made a good choice for waiting for the dfi nf3 agp board, which should be due out in april as well ,yippeeee :banana: :banana: :banana:
prices will probably be the same unless stores start price gouging us....
Well if all you guys pre-order or make a mad dash for them on the first day they are available the whole supply/demand thing will kick in and they'll "gouge" us. Just sit on your credit cards a few weeks/months and then when I give the word ... we all upgrade. ;)Quote:
Originally Posted by MaxxxRacer
Quote:
Originally Posted by TEDY
time to find a new job, min wage here is $5.5 but people cant buy a car, house, whatever at less than $20 hour.
I 2nd that :toast: although yes I realize this thread is almost a month old now haha.Quote:
im really excieted about this, now i will feel like i made a good choice for waiting for the dfi nf3 agp board, which should be due out in april as well ,yippeeee
For anyone who actually does perhaps come back to this, is it basically being agreed upon then that the San Diego will actually be the core to get? For not only the extra cache but the fact it should OC better too? But of course this is all speculation but was just seeing what the general concensus is as of right now :p
Nothing about the FX-55?
That would mean that we have to wait until the FX-57 for 90nm; it could mean that this new revision isn't that great.
there is no 90nm fx55, its not on the roadmaps. The Diego will be 4000+, 4200+ and fx57. NO mention of fx55, NO point when there already is a fx55, amd only makes one line of fx at a time and discountinues the older one
That....or this particular silicon design and architecture isnt meant for the power that those higher chips posess. The winchesters were supposedly not that great to go above 2.2GHz (obviously this isnt the case as many people have, just quoting AMD release article in regards to venice) and that the venice can hit 2.8GHz fairly easily but that 2.8GHz is somewhat a cap off point for the silicon architecture.....
So simply put I think it's just mainly meant for these lessor chips not a server processing power kind of chip like the FX series.
Don't know this is all my speculation and ranting ;)