macci
My friend had an unlocked Palomino which WOULD NOT do 200 FSB @ whatever multiplier. Maybe an isolated incident, though.
ALL
Btw, who says Dothan has a _short_ pipeline? 15 stages at least.
Printable View
macci
My friend had an unlocked Palomino which WOULD NOT do 200 FSB @ whatever multiplier. Maybe an isolated incident, though.
ALL
Btw, who says Dothan has a _short_ pipeline? 15 stages at least.
Guys, stop all the speculation. Those of us who HAVE Dothans see that mem b/w makes a HUGE difference with them. These chips are strangled and would be FX - competition with more mem b/w. Even in 3dmark 2001, the Dothan puts up wicked system based scores...
Like I said earlier, a simple chipset change (like Granite Bay of lore) would make the difference.
Well, thats a P4 for you. :DQuote:
Originally Posted by macci
I've heard... and often... that the P4EE cache is not inclusive, and that its actual maximum cache capacity is 2.0MB. I'll look for a link sometime.
Meanwhile you can search for a non-cursed FX-55. ;)
*edit*
Eintausend!
didn't mention anything bout die size at all. As for fsb, as you can clearly see from increase in fsb the cpu works more efficiently, always has and will do as it can flood the cpu with work in basic terms. The issue was that under 200Mhz fsb i believe the cpu that was destined to be 20W when it was released ran at 27W or something and wasn't entirely stable, which should tell you something(remember this is on release of dothan which is a while ago now and was probably the main issue being worked on).
AS for fsb being determined by the cpu, its a case of how a cpu works rather than anything else. as for bandwidth being needed, as with most cpu's thats the case, until we see dual channel setup no one can really say what will happen. AS i tried to hint at, even a ath xp which isn't as good as the dothan got not a huge, but a noticeable boost in benchmarks upping fsb to a 133/166 ratio. Because even on the best chipset basic efficiency of a mem controller can't ever be 100%, worlds not like that, on a ath xp the extra speed on the mem while messing with the pattern of communication just caused it all to work better together. For all we know its simply a case of that, or one of many possibilities. We will have to see if/when it happens.
As for centrino, well you can pretty much put one guy on ensuring people remember to add wireless into the laptop, everyone else works on the chipset and cpu's- most of them being on cpu. IIRC the team is around a 1/3 the size of the p4 team. P4 design is a trial and error design, they literally spend most of the time in design moving transistor by transistor till it all works together, as their netburst architechture is so complex. i'll try to find a few good articles that explain that better than i did.
TBH i haven't read as much as i have on dothan as other cpu's, for instance, is the connection from cache to cpu 4x fsb style connection, or is it mem to cache? from what i understand for a cpu to be this low the actual processing parts, none cache parts have to be unbelieveable simple to not waste energy, which would make cache, amount and efficiency of cache/chipset contact so important.
ANother thing that pops up in my speculation radar is this. Intel are moving to a advertising stratagy of "more fsb rocks, who needs more Mhz anymore now we can't get anymore?" , if infact dothan could be that much better with higher fsb, wouldn't it make sense to whack up fsb right now on a chipset they already have so they can stretch out dothan clock bumps much further, make it last longer and make yeilds better.
True, but here's the deal. Intel has been pushing P4's on DELL, et al. forever now and HOW in the world will they come back and say, "Ermmmm.... we've decided to put our notebook chip into desktop rigs because it's a lot faster and consumes less power"Quote:
Originally Posted by drunkenmaster
Hehehe... the reaction would be, "You freekin' morons... then wtf have you been pushing on us for the past few years?"
c
Exactly.....that's the main reason why Intel is not bringing the Dothan to desktop. They've put so much money and time for the Dothan into being a mobile chip that they don't want to mess that all up by bringing it to the desktop. It's all about the centrino technology.....I mean how many people actually know that centrino is the wireless technology and not the processor? People would go from 800fsb high speed P4's to low speed lower FSB Dothan and it's faster and more money? I think Intel is just trying to fool us......Quote:
Originally Posted by charlie
yup, theres no way intel can back away from p4 till lets see, basically they've just forced everyone mobo manu's, mem makers, oem's, the dells to go towards 775, ddr2 and prescott. THey can't kill it that quickly, thats why that will be there main thing till at least a year after 775 was available, its just the way stuff works. Dothan will have to go ddr2 and ramp up to equal fsb's, chipset functionality as p4 stuff. Thats why prescott will go dual core because they will have to too be able to compete with amd dual core stuff, which extends p4 even further. 2006 till dothan dual core main cpu stuff IMHO.
But still, right now it would be cheaper and easier for intel to not release new cores of dothan and ramp up fsb, if they could, IE instead of a smaller yeild 2.4Ghz they could keep it at 2Ghz and go to 200Mhz, then 266Mhz, and ddr2 and so on, it would be cheaper, bolster their "fsb rocks Mhz doesn't" new slant on things, and increase the time they can release the dothan speed bumps in, can't see a reason not to do it unless there are some kinda issues, which were around when first dothans came out.
my whole point here(just going by neo's response and what i think people think i'm saying) is that ignoring dothan going to desktop, increasing fsb and keeping lower speeds would make perfect sense right now for intel, so you have to think why they aren't.
Thing is, the p-M's power was clearly known by intel before 775 stuff was released. Even at lower fsb's canceling 775 stuff, waiting a few extra months and starting to get P-M at 2Ghz speeds into desktops would have made more sense than commiting themselves to a overheating, motherboard stressing complete product cycle, someone made a big boo boo there- BIG. Not only are they having to keep people working on increasing p4 speed, but they've now absolutely got to go dual core prescott too, which is just ridiculous.
We all know what they should do but it really sux that we cannot do anything about it. Just be happy that we have the knowledge to know what is good and what is worth it's money. I know that I won't be buying an Intel processor for a long time with all the garbage they are releasing now....except if they change what they are doing now and work on the Dothan....then I might think about it.Quote:
Originally Posted by drunkenmaster
There's a thread with an Alviso-chipset Dothan board in the Intel section. It looks really awesome - DDR2 + PCI-E. I just wish we had some idea of how long it would be before they come into production.
Well they delayed it til early 2005 or so they say. They wanted them to come out in 2004 but problems arose and yada yada yada. DDR2 and PCI-E are a must because everyone is now seeing DDR2 and PCI-E and say well why can't my laptop have those features. WHY AM I GETTING RIPPED OFF? So just hopefully they'll come out soon and then we can see maybe some well deserved life out of these dothans.Quote:
Originally Posted by Shade00
Hallowed
Actually, the effective cache size is 2 - .5 so around 1.5 mb. It is inclusive.
Shade00
What's so awesome about DDR-II?
Ivan: DDR2 should provide some tangible bandwidth increases. I'm also pleased to see that PCI-E will be available for graphics.
Shade00
TCCD @ "600" 7-3-3-2.5 will own DDR-II even at "800" probably.
Well, I also hope that the Alviso chipset will bring higher FSB capability for the Dothans.
There is DDR2 that will do tight timings, i.e. 3-2-2, and that would definitely outperform the TCCD.
http://img153.exs.cx/img153/4091/axpvsdothan9oq.jpg
Is it just me, or does that dothan look like an AXP with a quad pumped fsb and SSE2? LOL! :p:
Oh the side effects of boredom.... :toast:
dippyskoodlez
Sandra is the dumbest benchmark in the world. In 3dmark2001, Dothan would slaughter Athlon XP at same clockspeeds.
If nobody was buying P4s, there might be some incentive for Intel to spend the money to develope the Dothan into a desktop platform. Cant see that happening though. Its all about the bottom line for Intel.
And best of all....
they OC well... even on AIR and a NorthBridge cooler......on the CPU. Try that with an A64, lol.....
edit, I mean the CPU is cooled with a northbridge cooler :D
Does your heatsink get hot? I wish I could put that chip in my laptop.
Quote:
Originally Posted by IvanAndreevich
I know, I just find it funny... same at each speed, just super pumped FSB...
Makes me wonder what an axp would do with a quad pumped fsb. :slobber:
dippyskoodlez
>>Makes me wonder what an axp would do with a quad pumped fsb.
Then look at A64 - HyperTransport is a better bus than Intel's.
2500+ #1 - Ran extremely stable at 10*200MHz with a core voltage of 1.7v.Quote:
Originally Posted by macci
2500+ #2 - Ran stable at 10*200MHz with a core voltage of 1.85v. Notice I did not use he word "extremely". If left running for a period of 15 to 18 hours, the system would finally crash to the desktop.
2500+ #3 - Would not run 200MHz FSB at all, no matter the voltage
linkie : http://www.hardocp.com/article.html?art=NDUz
does this mean the maxmem 104 tweak does more bad then good to a P4 or dothan system?Quote:
Originally Posted by macci
charlie, what's the max you have gotten that chip Prime stable? Also, what about the max fsb on that new board? :confused:
Well, I don't DO Prime95, just don't care I guess... but it does 2711 sPi 1M stable and 2741 piFast stable. And I still get the same 160+ errors as before, even using agp/pci up to 76/38.... you know, the SLOW MOTION effect in Windows :(Quote:
Originally Posted by WiCKeD
C
a7n8x is one of the worst nF2 boards ever made. :stick:Quote:
Originally Posted by kristos
Whats the ram mhz got to do with maxmem tweak?Quote:
does this mean the maxmem 104 tweak does more bad then good to a P4 or dothan system?
charlie
Slow motion effect? Can you tell me about that please.
well, if you double click a desktop icon, it takes 7-8 seconds to appear. Sometimes you won't see sPi tabulating results and then 3 or 4 lines will appear. When you run the cursor over the start button, the START disappears, when you push the button, the start box opens in 5-10 seconds.....Quote:
Originally Posted by IvanAndreevich
slow-motion effect :D
Probably related to Display Driver
charlie
Did you try some el-cheapo ancient PCI videocard?
no, but I tried the onboard VIDEO and it does the same :(Quote:
Originally Posted by IvanAndreevich
kristos & co. AXP, okey well maybe some of the AXPs have some wierd problems running high FSB. But thats AMD and AXP not Intel and Dothan :D
Also the link w/ 3 AXP2500s FSB test was hardocp OC-article so pretty much anything is possible ;)
charlie, 160+ should work better if you use the integrated video (without driver)?
edit: nevermind didnt' see your last post :D
charlie
The thing is, the onboard video is AGP as well. I would try PCI :) BTW, when you try PCI you might also try killing the AGP driver in the device manager.
I can see why it may have problems with AGP drivers, since this was intended as a laptop chipset.
macci
>>OC-article so pretty much anything is possible
Yes we all know [s] and their vast knowledge of hardware.
but that doesn't change that the only thing different on that setup were the processors so the difference in FSB can (should) only have been due to the processors themselves.Quote:
Originally Posted by STEvil
I thought it was the amount of memory used set in with maxmem but the mhz clock seems more logical :D my badQuote:
Originally Posted by STEvil
is there any review site that you mad clockers do approve?!? :pQuote:
Originally Posted by macci
I mean tomshardwareguide ain't good, hardocp ain't good, anandtech is better but still not good enough...
where does a guy like me have to get his info then :confused:
Each processor required a different voltage to reach the overclock they were doing, so that may have affected the power being delivered to the rest of the system.. that indicates a PSU or motherboard issue (motherboard more so since FSB suffered).
What they really should have done is tested each on two or three boards known to do both high cpu clocks and high FSB's, as well as tested the CPU's on say an older Epox 8K3A+ to be sure it wasnt just an nF2 issue..
But thats getting into a lot of work to do..
kristos
>>is there any review site that you mad clockers do approve?!? :p
FORUMS. People get new hardware here and they write about it. Better than any review out there ;) Because -
1) They will listen to what you have to say and respond
2) They will do their best to resolve any problems because they usually pay for it with their own money, whereas reviewers will only :banana::banana::banana::banana::banana: and cry that something doesn't work ;)
[Sorry for getting offtopic]
They don't appear to love PiFast, though... like ALL Intel's.
Anything interesting to conclude the year with?
i know that 3dmark03/05 are more GPU-friendly and not so CPU-friendly, but has anyone run Dothan on these benchies?
Nope....Quote:
Originally Posted by IvanAndreevich
Already turned my attentions to other CPU/MOBO combinations :D
C
so you're just saying every reviewer out there does no effort to do any experiments with overclocking or finding out what's wrong and trying to fix it?Quote:
Originally Posted by IvanAndreevich
DFI is releasing a second revision of the 855GME with voltage control and locks fixed.
This board will also feature mounting holes for a aftermarket socket 478 HSF
Sounds good :cool:Quote:
Originally Posted by FUGGER
Might could be something good there FUGGER :)
when will this be? my DFI 855 is still in RMA, if they plan on bringing out a new revision soon, I might be able to arrange something with the shop where I bought it from (if they plan on replacing my board that is).
FUGGAR can you tell us anythig about your source of this information, links etc......?
Here is one - notice GT1 ;)Quote:
i know that 3dmark03/05 are more GPU-friendly and not so CPU-friendly, but has anyone run Dothan on these benchies?
Amazing score...at gt1 i have 426 with 3.4 EE at 4.6Ghz!!!
macci, dang....that's uber......and tnx for answering and giving the link =)
jmke
Not referring to madshrimps ;) Obviously you guys are cool to even hang out here.
it depends on the target audience of the website also, quite a lot even; for example a CPU stepping for Joe Shmoe doesn't mean a thing afaik
readers of TomsHW are basically those people, sometimes they do venture a bit further but they try to keep it noob friendly,
[H] is in the middle of the TomsHW and Extreem OC sites, trying to balance crazy overclocks, with valuable data for Joe Shmoe in their reviews, sometimes they succeed, sometimes they don't. But simply put, it's 99% of the time due to time constraints that not every aspect can be explored to the roots.
Whereas one overclocker could experiment during weeks with one motherboard, a reviewer has to check out 5 of them, besides a vidcard roundup and a heatsink test, if you know what I mean:)
while the info he will provide in his review on the motherboards will almost never satisfy the overclocker completely, it will pretty much be more then needed for the average HW website reader and Googler
I saw the new 855GME at CES in the DFI meeting room.
FUGGER
Visually any different significantly?
Macci, what have you done to the X800XTPE to get those clocks? I have a fresh one here and an prepping to mod it, looking for insight from experts.
wow, any "visual" impression from your side? any thing significantly different visually? like more caps, or mosfers riddles with heatsinks, etc?Quote:
Originally Posted by FUGGER
any idea of wen this is cumin out mate? i was jus about 2 order the Aopen aswell..... :DQuote:
Originally Posted by FUGGER
For those of you interested pics and specs of AOpen's new i915GMm-HFS board can be found at these sites:
http://www.vr-zone.com.sg/?i=1659&s=1
http://babelfish.altavista.com/babel...15gmm-hfs.html
http://babelfish.altavista.com/babel...15gmm-hfs.html
Looks like we're looking at March availability for this one. Can't wait.
Mmmmmmm.....
makes me want to KEEP my FS-735 CPU.....
Man, this could be awesome. Now, if they would just do dual-channel...:slobber:Quote:
Originally Posted by FUGGER
Doesn't the sonoma chipset do dual channel?
yes it does. only in DDR2 or something like that... i dont remember exactly
yes, the new aopen can do dual channel. but only with ddr2 and not with ddr1.Quote:
Originally Posted by grenadier
Well if the lock issue is really fixed on the DFI, then these things should smash all with phase, even with single channel. :slobber:
but how many secs is a dothan@3300mhz???
For some reason Dothan sucks arse at 32M while dominating 1M..any ideas?
Dothan running at 2.0GHz on an AOpem 855 desktop board Mushkin PC3200 Level 2 @ 167Mhz, 2-2-2-5. WinXP sp2.
32M = 43:46
1M = 40
Winch 3000 running at 2.0Ghz on Epox 9NDA3J Ballistix@222Mhz 2-2-2-7 WinXP sp2
32M = 35:22
1M = 42
A full eight and a half minutes faster on A64!!.. Maybe it's just one more of Dothans collosal performance flaws???? :confused:
DFI's new board.
http://www.xtremesystems.org/forums/...id=23016&stc=1
I have word Asus maybe doing a board also thats 915 based with Dual ddr1 support.
can't wait :)
looks good :DQuote:
Originally Posted by bigtoe
Doesn't the mounting system on that DFI board look like lga775? I thought it was going to be 478?
in 1M the dothan is faster, because of its 2mb l2-cache, so the cpu didn't have to use the memory much. in 32M/16M it has not enough bandwith to beat an amd64.
Again, it's ALLLL about memory b/w with the little Dothan... this next-gen of Dothan mobo's will hopefully bring them front and center!!
C
Will that fix these problems?Quote:
Originally Posted by charlie
http://images.anandtech.com/graphs/l...91256/5839.png
http://images.anandtech.com/graphs/l...91256/5861.png
And it's sorry performance in things like CAD, Mathmatica, Winrar...?
Dothan just seems not well rounded...not ready for prime time on desktop.. also I notice it's not very "snappy" if that the right word like A64.
All due to memory right?
what did they change? what chipset? :slobber:Quote:
Originally Posted by bigtoe
Chipset is 915:)
thats the same as the new desktop chipsets right? damn so no more FSB limitation :D
oc'ed dothan 2.0 + sp-94 + panaflo l1a undervolted = :slobber: :banana:
I want to see what one of these will do under Phase, or very good watercooling.
With better on-board features, we could be looking at the next great overclocker's chip. From what I have read, it is all about the motherboard at this point.
Better motherboards = better memory performance/limits. Better memory performance/limits = better across-the-board performance
Heck you might not even need the panaflo in a well vented case.:DQuote:
Originally Posted by Supertim0r
That's what I'm most interested in about Dothan.. Building a truley silent SSF box that can still perform. Need to fiqure out something with Vcards though.:)
That BFG I have you can literally hear in the next room.:(
what is wrong with this picture?, i see 2 ide channels and no second controller :confused:Quote:
Originally Posted by bigtoe
I think Charlie's talking about jumping from SC FSB100/133 to DC FSB200, which would accelerate the Dothan quite dramatically - just look at how it reacts to going from 1:1 to 3:4 :)Quote:
Originally Posted by Zebo
Uhm, on the picture it says 852/855 GME on the right side of the socket... so this isn't a DFI with 915 chipset but the pro moddel of DFI's i855 board, the one fugger reported about earlier, right?Quote:
Originally Posted by bigtoe
right,
oh and bigtoe... any chance to get a bit more info on that asus 915 with dc-ddr1? that sounds like a true winner right there.
I cant wait for dual channel on the Dothan. We need a new King of the Hill.
:D Yep - will be a nice one, but does it work like this:Quote:
Originally Posted by Franky 4 Finger
http://de.geocities.com/hardcorecloc...bench15403.JPG
:toast:
i made a post about this in the news section and it disappered :(
sonoma is better than nforce4, not sure how AMR compares
but in aust, 400 for 725 dothan 1.7g and another 400 for the aopen. god knows how much the new one wil be
Really? :hehe:Quote:
...but reconfiguring the cpu die to accept higher fsb, more bandwidth and so on isn't something that can happen instantly.
There are documented issues that stopped the release of speeds of dothans initially as they couldn't get certain fsb's working, i can't remember exactly but i think the dothan was supposed to be a 200Mhz fsb part, but it simply wouldn't work at that, in single channel, so was released at a lower fsb. SO there are issues there.
http://u-san.net/c-board/file/M760-ddr266-fsb341.gif
(by TAM)
Nothing wrong w/ Dothan and high FSB :D
hmmm thats odd, i thougt those chips can only do 200fsb max? :confused:
is this a new revision or was the chipset limiting after all and not the cpu interface?
The original Dothan mobos (like DFIs i855 and AOpen i855) couldn't run high FSB.
Now with the adapter and i865/875 mobos things seem to be totally different.
ahhh so it was the chipset after all! good to know :)
i should get one of those boards with adapter then and see what my old 1.4ghz banias can do ^^
it survived all those tests with my homebuilt pentium m adapter that i never managed to finish ^^
time to let it have some fun after all those months of torture :lol:
I figure this might be the thread to ask the following question, what is the average OC for dothan's on air, and does initial speed make much of a difference. The reason I ask is people have found a way to OC dell 9300 notebooks, by grounding a pin on the 400 mhz fsb variants which allows:
1.5 ghz to 2.0 ghz
1.6 ghz to 2.13 ghz
1.7 ghz to 2.26 ghz
1.8 ghz to 2.4 ghz
since you can only go to one speed (this is partly true) you must be stable at the higher speed. Anyway people have had very good results up to 2.13 ghz with some success at 2.26 ghz, but 2.26 ghz and 2.4 ghz is not very consistent....I'm just wondering if this good be a PSU issue or is it necessarily a processor issue?
Or a heat issue...after all it is in a confined notebook.Quote:
Originally Posted by socrilles