To all who trash conroe

Status
Not open for further replies.
duby229 said:
It goes back to that old saying "The truth, the whole truth, and nothing but the truth"

What you say is true, but it isnt the whole truth. You fail to take into account interupts and so on, which still to this day account for quite a bit of bandwidth. There are still others as well. PCI, and hence forth also PCIe addresses need to be addressed byt the CPU, and just a ton of other things as well. It goes a lot deeper then I can explain. That is the bottleneck.
Click to expand...

Yup and I said that was the simple version minus buffers, ECC and etc... Please don't bring up the etc.. as if that's not what I was talking about. Interupts are part of the DMI, as is I/O, PCI BUS and etc.. on Intel's system as well, that's why I said look it up? Now there is a nice write up on APIC's if you like, go to Intel's site and read about it?

His explaination was spot on. What I think that what some people fail to realize is that memory addressing must be done by the CPU. No matter what device has that memory allocated. AMD uses something called an IOMMU, which to my knowledge has no equivalent in an Intel architecture, which means that all IO requests must be handled by the CPU.
Click to expand...

Don't cloud your already bad judgement with your dislike for me?

So If all Intel can use is DDR2-667, that it doesnt matter if the chjipset supports DDR2-800, it will be bottlenecked at lowerbandwidths.

That is what he was trying to say. DDR2-667 will saturate the FSB bandwidth. Therefore anything faster then that is simply wasted. It really doesnt matter if the chipset supports faster memory. All IO requests are serviced by the CPU anyway, so all that extra memory bandwidth is being wasted.
Click to expand...

I said 30 something posts ago that the Processor can't use more bandwidth than the FSB Provides. Now it's you guys who are drinking something for saying otherwise. If Conroe will be limited to its FSB 1066 or 8GB, plain and simply without a 1000 word post, and Dual Channel DDR2 800 that provides 12.8GB is used, the rest of the system will be abled to use some of that unused bandwidth.

One more time, ALL devices are NOT running on the FSB and are NOT COMPLETELY controlled by the CPU.

Intel is in the same situation that the nForce 2, and Athlon XP was several years ago. The chipset provides mor memory bandwidth, then the FSB can funnel to the CPU, and it is beingwasted.
Click to expand...

Please put down the Crack pipe and Back away slowly? Advanced Buffers and DMI as Intel uses it weren't in play then. nVidia's Dual Channel was setup for its IGP, not the AthlonXP.

All this chatter and why, because Intel announces support for DDR2-800 for Conroe and you guys are trying to say its a waste, Gimmick or etc... right? No, that's trashing Conroe and very much the subject what this thread was about. I didn't say it would improve Processor Performance, I said SYSTEM PERFORMANCE.
 
duby229 said:
Intel has to funnel Memory bandwidth through the FSB. Therefore ONLY FSB badwidth is meaningful.

AMD can access both the Memory and the HTT at the same time, so they are additive badwidth. (aggregate if you prefer)
Click to expand...


This sounds like BS to me you have any documentation to back this up.

I fail to understand how everything is not routed throught the htt just as fsb with off-die memory controller.

edit : nvm i foudn it myself
Intel
MC1.gif

AMD
MC2.gif


Maybe Intel is saving IMC for an ace up it's sleeve??
 
nForce2 was optimized for its Intergrated Graphics Processor not the AthlonXP.

http://www.viperlair.com/reviews/cpu_mobo/retired/s462/7nif2/index.shtml

Viperlair said:
We can see that in the CPU test the nForce 2 doesn't show any advantage at all compared to the SiS 745 chipset. Apart from the single channel mode IGP the systems all have extremely similar graphs, with only a few small differences between the different cards/platforms. The differences between the different systems is small to non-existent differences between the SiS 745 and the nForce 2, in either single or dual channel mode. In UT2003 the difference between with IGP enabled and it disabled is pretty high, about 25% in dual channel mode, though this may be more a video card difference than that of the lowered bandwidth to the CPU. However the differences between single channel mode and dual channel mode are very high, about 40.4%, which is nice to see. How about when we increase the resolution to 1024*768?
Click to expand...

82975X Memory Controller Hub
http://www.anandtech.com/mb/showdoc.aspx?i=2631
 
Conroe is a beast I think, well I'm just going to wait till I can get 2x more performance till I upgrade, anyways my [email protected] only gets 11,500 on the Aquamark CPU test.
 
LoL. You guys are soooo focused on FSB's and bandwith that you certainly miss some major points.

1) DDR2 800 instead of 666 can improve Conroe performance. Why? Cause DDR2 800 can have better latency - and that's what matters both for Intel and AMD.

2) Don't forget Anand's article about Athlon64 E core and unnoficial DDR500 support. And how tiny the gains are? Even if synthetic test bandwith goes up, that doesn't translate in the increased real world performance. (cause real world app makes use of cache subsystem and has to actually chew data)

3) Is Intel's Conroe 1066 FSB not enough for today's apps? Hard to say atm.
Best way to increase memory subsystem perfomance is to not use it at all. It doesn't take a rocket science to compare L2 cache access time and memory access time - penalty for L2 cache miss is brutal for both Intel and AMD. Unless Intel has incredibly stupid L2 management mode, Intel will have more cache hits. BUT from what i have seen on the web - Conroe is some sort of execution monster so there could arise some scenarios ( larger pressure on cache due to 64bit mode etc etc etc, both cores chewing truckload of data and competing for L2) where bandwith limits of FSB will kick in. (Tho in Intel's credit i have to say, that in most of such scenarios AMD will probably take even harder hit due to no cache sharing and having 2 cores with tiny amounts of L2 sitting on a single IMC)

4) Having memory controller in chipset has advantages if you have integrated graphics or truckload of DMA going on in system. Extra memory bandwith is usefull in such situations. And integrated graphics are without a doubt very important to Intel's business.

5) K8 vs Conroe book is already closed. Clock for clock K8 is badly owned. Period. If you don't agree - look around the web - Conroe is beating K8 in every possible way.

6) AMD has to either increase clock or to release K8L to compete with Conroe.

7) AMD has problems with AM2 in desktop. No self respecting Conroe (both performance and pricing) aware buyer would consider AM2 at current pricing levels - no performance gains and Conroe is out real soon.

8) If we make assumption that AMD will release K8L this early autumn and it will be able to beat Conroe, we should consider that Intel is not sleeping either. I have no doubts that Intel is working 24/7 on improved version of Conroe for 45nm tech or even possible 65nm revision
 
Nice post DariusD . The only thing i don't agree with is K8L beating conroe. For AMD to come up with a 30 to 40% performance gain in a short time just doesn't pass by me. I wish the Bookies took bets on this as I would go all in against AMD pulling that off in such a short time.

If the 2.1ghzConroe O/C to 3.1 simply stumbles there because of FSB headroom and lack of Multipliers which I believe to be the case. Than a 2.66 conroe = an easy 3.33GHz O/C
 
$BangforThe$ said:
Nice post DariusD . The only thing i don't agree with is K8L beating conroe. For AMD to come up with a 30 to 40% performance gain in a short time just doesn't pass by me. I wish the Bookies took bets on this as I would go all in against AMD pulling that off in such a short time.
Click to expand...

We have exactly zero hard data about K8L. Only rumours about more floating point execution units and core redesign. It's definately hard to gain 30-40% more performance per clock without a major architecture redesign ( even most floating point execution units in the world won't help if you can't keep them well fed -> Intel has learned lesson with 2 dual pumped ALUs - without beeing able to feed them, you are just wasting die space).

And about bets - imho AMD will rise clock on 65nm tech as much as possible and will introduce ~10-15% average per clock improved K8L ( improvement vs K8). What i really wonder - is what clock headroom Conroe has on 65nm tech? Intel has been talking about 3.33G EE chips so early in Conroe/65nm process life that one has to start dreaming about 4G oced or not Conroe chips after new core rev comes out.
 
Time will definitly tell which one will come out the victor in the Conroe-K8L battle, although I'm not going to build a new system till next year so then I can get all next gen stuff.
 
sculelos said:
Time will definitly tell which one will come out the victor in the Conroe-K8L battle, although I'm not going to build a new system till next year so then I can get all next gen stuff.
Click to expand...

If my wallet is OK, I'm going with E6600 Conroe/i965 soon as they launch. Don't like doing anything early but I have a feeling the VARs and OEM will drive the prices up like they did with DC Sc-939 Opterons that Sparks and I were shopping for.
 
DariusD said:
LoL. You guys are soooo focused on FSB's and bandwith that you certainly miss some major points.

1) DDR2 800 instead of 666 can improve Conroe performance. Why? Cause DDR2 800 can have better latency - and that's what matters both for Intel and AMD.
Click to expand...

Actually, if you look, that's exactly what I said - that Conroe would benefit slightly simply due to DDRII-800 being faster. It won't, however, benefit from the additional bandwidth.

2) Don't forget Anand's article about Athlon64 E core and unnoficial DDR500 support. And how tiny the gains are? Even if synthetic test bandwith goes up, that doesn't translate in the increased real world performance. (cause real world app makes use of cache subsystem and has to actually chew data)
Click to expand...

And I pointed to this as well, stating that DDRII-800 would only allow a very slight performance increase over DDRII-667.

3) Is Intel's Conroe 1066 FSB not enough for today's apps? Hard to say atm.
Best way to increase memory subsystem perfomance is to not use it at all. It doesn't take a rocket science to compare L2 cache access time and memory access time - penalty for L2 cache miss is brutal for both Intel and AMD. Unless Intel has incredibly stupid L2 management mode, Intel will have more cache hits. BUT from what i have seen on the web - Conroe is some sort of execution monster so there could arise some scenarios ( larger pressure on cache due to 64bit mode etc etc etc, both cores chewing truckload of data and competing for L2) where bandwith limits of FSB will kick in. (Tho in Intel's credit i have to say, that in most of such scenarios AMD will probably take even harder hit due to no cache sharing and having 2 cores with tiny amounts of L2 sitting on a single IMC)
Click to expand...

I think, that in terms of truly demanding multitasking scenarios, having each core with its own dedicated cache would prove the better scenario. My ideal situation is each core having its own dedicated cache, but then having an implementation of Smart Cache as well, so that the other core's cache could be accessed when it is absolutely required. But I do agree with you.

4) Having memory controller in chipset has advantages if you have integrated graphics or truckload of DMA going on in system. Extra memory bandwith is usefull in such situations. And integrated graphics are without a doubt very important to Intel's business.
Click to expand...

I fully agree as well.

5) K8 vs Conroe book is already closed. Clock for clock K8 is badly owned. Period. If you don't agree - look around the web - Conroe is beating K8 in every possible way.
Click to expand...

Agree again.

6) AMD has to either increase clock or to release K8L to compete with Conroe.
Click to expand...

Yep. Simply increasing clock speeds though will only cause the K8s to catch up some, and not truly compete I fear.

7) AMD has problems with AM2 in desktop. No self respecting Conroe (both performance and pricing) aware buyer would consider AM2 at current pricing levels - no performance gains and Conroe is out real soon.
Click to expand...

AMD *had* problems with AM2. From the latest previews, those look to be resolved. However, I do agree - at the moment, given the price of the X2 line on AM2 upon its release, there's no reason not to wait the extra month and a half and go with Conroe.

8) If we make assumption that AMD will release K8L this early autumn and it will be able to beat Conroe, we should consider that Intel is not sleeping either. I have no doubts that Intel is working 24/7 on improved version of Conroe for 45nm tech or even possible 65nm revision
Click to expand...

Well, K8L isn't slated till sometime next year (likely no earlier than mid-2007) at the earliest. Thus, I *highly* doubt there's any way they can push it out by the end of 2006.

Intel's plans up through 2008 are generally known (granted, those plans are subject to change), as they've published their roadmap for up through then, based upon the Conroe core. We'll see some single-core versions, the quad-cores will debut next year, and besides that, there aren't any major architectural changes from what we can see. However, it is early and obviously anything can happen and change.
 
$BangforThe$ said:
Nice post DariusD . The only thing i don't agree with is K8L beating conroe. For AMD to come up with a 30 to 40% performance gain in a short time just doesn't pass by me. I wish the Bookies took bets on this as I would go all in against AMD pulling that off in such a short time.

If the 2.1ghzConroe O/C to 3.1 simply stumbles there because of FSB headroom and lack of Multipliers which I believe to be the case. Than a 2.66 conroe = an easy 3.33GHz O/C
Click to expand...

Seen this?

http://www.dailytech.com/article.aspx?newsid=2034
 
Donnie27 said:
Seen this?

http://www.dailytech.com/article.aspx?newsid=2034
Click to expand...

No i hadn't . interesting . The comments are really funny . My Dothan @ 3.0ghz is good gamer better than amd but weak in other area's. I find it hard to believe that AMD beats intel by 20 to 30% in the p4 vs AMD64 battle I must have slept threw that. Hell other than a few apps. Multi thread. My wifes 3.2 C isn't 20 to 30% slower than AMD'S top of the line. You ever wonder were these people get this info. When you say something like that a link to back it up would be nice.
 
$BangforThe$ said:
No i hadn't . interesting . The comments are really funny . My Dothan @ 3.0ghz is good gamer better than amd but weak in other area's. I find it hard to believe that AMD beats intel by 20 to 30% in the p4 vs AMD64 battle I must have slept threw that. Hell other than a few apps. Multi thread. My wifes 3.2 C isn't 20 to 30% slower than AMD'S top of the line. You ever wonder were these people get this info. When you say something like that a link to back it up would be nice.
Click to expand...

I think it will come down to them not having their Shee-yte together and more shortages like the last two quarters of 05. If that's really their view, then they're pretty lame to say the least.

I find it hard to believe as well, because they don't LOL!
 
ToastMaster said:
I think, that in terms of truly demanding multitasking scenarios, having each core with its own dedicated cache would prove the better scenario. My ideal situation is each core having its own dedicated cache, but then having an implementation of Smart Cache as well, so that the other core's cache could be accessed when it is absolutely required. But I do agree with you.
Click to expand...

It certainly goes beyond the topic of this thread, but ppl fail to realise that unified cache is HUGE step ahead. We live in the world where probably >95% of usage scenarios are single threaded - so having unified cache is already a win.
Now the rest 5% of multithreading is based on paradigm of "data sharing between threads is bad for your app health" - bus, cache sinchronization, NUMA, Microsoft and hell knows what else will kill your performance. You usually have threads working on either different "tasks" or different chunks of data. ofc having faster "communication/sharing" will help, but effect is tiny ( and tiny is overstatement probably).

And app devs cant run to start using those low latency data sharing mechanisms either - multiprocessor etc systems will thoroughly penalise them for doing so.
 
Wow, this conroe chip is going to be insane. I hope the price isnt going to rape my wallet.
 
comedy said:
Wow, this conroe chip is going to be insane. I hope the price isnt going to rape my wallet.
Click to expand...
Prices have already been released bro.
 
PerfectCr said:
Prices have already been released bro.
Click to expand...

No shit? I havent looked through this thread thoroughly, what is the price?

Edit: nvm, just found the prices on the theinquirer, that is very cheap, these chips are gonna sell like hotcakes. I hope its not gonna be sold out everywhere like when the 7900gt was released :D
 
Yes the yonah @ 3.66GHz was super cooled. Most of the talk is about the Conroe running at 3.1 ghz on air . a 50% O/C . The M/B seems to have run out of Bus speed overhead to allow higher O/C . Cpu multi. could have had something to do with that also. If a 2.4 Conroe can achieve the same O/C on air with that M/B we are talking about a 3.4 ghz Conroe that will beat everthing out there by 50% . So a stock Conroe @ 3.33Ghz is going to be very hard for AMD to overcome anytime soon .
 
I know many of us say we like intel for consistances and such. When I play online games I can usually tell if its an AMD or intel server. When theirs alot of errors going on in game play 99% of the time you can bet your playing threw a AMD server.

This just released info. Notice how AMD plays down the number of bad dies. This won't help AMD secure any US government sales.

http://www.theinquirer.net/?article=31354
 
$BangforThe$ said:
When I play online games I can usually tell if its an AMD or intel server. When theirs alot of errors going on in game play 99% of the time you can bet your playing threw a AMD server.
Click to expand...
:rolleyes: Ridiculous.
$BangforThe$ said:
http://www.theinquirer.net/?article=31354
This won't help AMD secure any US government sales.
Click to expand...
Yeah, I don't think that's any good for anyone. No one, Intel included, likes to do recalls. Still, I'm happy to see that this is such a tiny number of bad chips. When you ship litterally millions of units, things can happen. Here's the summary of your link:
Advanced Micro Devices is trying to track down and replace as many as 3,000 Opteron processors that could produce "inconsistent results" under an unusual, high-temperature circumstance.
Click to expand...
 
I read that report a little differant. The reason the chips got thro because of a test process weren't good enough and amd just caught that. So if the test wasn't good enough to catch all the possiable bad cores. That could mean a lot more than 3000 cores are involved. AMD would naturally try to downplay this . lets see how it unfolds.
 
$BangforThe$ said:
I read that report a little differant. The reason the chips got thro because of a test process weren't good enough and amd just caught that. So if the test wasn't good enough to catch all the possiable bad cores. That could mean a lot more than 3000 cores are involved. AMD would naturally try to downplay this . lets see how it unfolds.
Click to expand...

I really hate to be the one to remind you of all the Intel recalls......

Shit happens. Unfortunately it was AMD this time.
 
duby229 said:
I really hate to be the one to remind you of all the Intel recalls......

Shit happens. Unfortunately it was AMD this time.
Click to expand...

Where you not the gut that claimed that Yohan couldn't be OD'ed? :)

http://www.xtremesystems.org/forums/showpost.php?p=1419993&postcount=2021

s7e9h3n> So is your cpu NOT FSB limited?
s7e9h3n> Are these results basically due to the ACTUAL physical limit of the cpu?
--
My CPU is FSB limited at around 282MHz...
It can boot up with FSB=280 BIOS setting under DryIce or LN2 cooling.
In my case, there is no difference for FSB limit between DryIce and LN2,
and DryIce or LN2 gives about +15MHz FSB gain compared to water cooling.
Click to expand...

Terra - looks like you where mistaken?
 
$BangforThe$ said:
I read that report a little differant. The reason the chips got thro because of a test process weren't good enough and amd just caught that. So if the test wasn't good enough to catch all the possiable bad cores. That could mean a lot more than 3000 cores are involved. AMD would naturally try to downplay this . lets see how it unfolds.
Click to expand...

http://www.hardforum.com/showpost.php?p=1029290567&postcount=202

Techweb

Because of the tests, AMD has changed the screening process for rating the two product lines as the chips come off the production line, Taylor said. As a result, some chips that would have been rated with clock speeds of 2.8 MHz in the past would be listed at 2.6 MHz, making them less likely to be used in extreme computing environments.
Click to expand...

;)
 
So what's all this mumble jumble about Conroe :p somebody mentioned about 2.4 "on air" ? Any clue as to what that term means?
Is it really worth waiting till this fall or winter to pick a machine up ? or should one build a system and then ponder upgrading later?
 
Fort_Major said:
So what's all this mumble jumble about Conroe :p somebody mentioned about 2.4 "on air" ? Any clue as to what that term means?
Is it really worth waiting till this fall or winter to pick a machine up ? or should one build a system and then ponder upgrading later?
Click to expand...

Depends on what your current system is, how satified you are with it, and how long you plan to keep the system, if your current system is adequete and you won't upgrade for a couple years I think conroe would be worth waiting for, however if you upgrade every year then I would just build a nice system now and upgrade later.
 
Fort_Major said:
So what's all this mumble jumble about Conroe :p somebody mentioned about 2.4 "on air" ? Any clue as to what that term means?
Is it really worth waiting till this fall or winter to pick a machine up ? or should one build a system and then ponder upgrading later?
Click to expand...


I would build a AMD system now if I were YOU. Conroe is released in july . Someone like yourself probabablely couldn't buy a conroe till like oct. nov. maybe dec. So if I were YOU I would build a DDR1 AMD now. :rolleyes:
 
I am on an AMD XP 1800 with 1.5GB of ram currently. I have an AMD 64 3400 at work. I am not extremely happy with the XP 1800. 2-3 years of BSODs. Finally the system does not BSOD anymore. The 64 at work. Between the matrox video card on it and the chip I greatly dislike it.

I'm going to go Intel this next system, and I'm merely wondering whether I can hold out for another 1/2 a year on this system or whether I should just build a system now.

Any clue what the term "2.4 on air" meant ?
 
$BangforThe$ said:
I would build a AMD system now if I were YOU. Conroe is released in july . Someone like yourself probabablely couldn't buy a conroe till like oct. nov. maybe dec. So if I were YOU I would build a DDR1 AMD now. :rolleyes:
Click to expand...
What do you mean by "somebody like yourself" ?
 
2.4ghz core using fan to cool heat sink. If you believe that Conroe can't be had before oct. nov. Dec . as you posted. you would be a pesimist. These kinda of people should own AMD cpu's . If you bought an Intel it would change your life and make you become a an optimist. Now you wouldn't want your life to change just because of a CPU would you?
 
f_a_nboyism really irks me. Brand loyalty, whats the point? Buy the best performance for price that you can afford, be that AMD, Intel, ATi, nVidia.

I went from Intel/ATi (P4 1.8"a" @ 3.2 && 9600XT) to AMD/nVidia (3800X2 && BFG 7800GTX).

Reason? Price for performance, at the time that I bought. the 7800GTX was about 580$ CAD, compared to the X1800XTX's wopping 750$ CAD. :rolleyes: Pretty easy choice. And with intel motherboards as expensive as they are, it was again an easy choice to go with AMD.

If I were to build a system when conroe was out, with it showing to be the clear performance king at my price level? I'd buy conroe.

Buy what works, down with brand loyalty.

-Cameron
 
TrueChaos said:
f_a_nboyism really irks me. Brand loyalty, whats the point? Buy the best performance for price that you can afford, be that AMD, Intel, ATi, nVidia.

I went from Intel/ATi (P4 1.8"a" @ 3.2 && 9600XT) to AMD/nVidia (3800X2 && BFG 7800GTX).

Reason? Price for performance, at the time that I bought. the 7800GTX was about 580$ CAD, compared to the X1800XTX's wopping 750$ CAD. :rolleyes: Pretty easy choice. And with intel motherboards as expensive as they are, it was again an easy choice to go with AMD.

If I were to build a system when conroe was out, with it showing to be the clear performance king at my price level? I'd buy conroe.

Buy what works, down with brand loyalty.



-Cameron
Click to expand...

Interesting post. When did you buy that 7800GTX and Where were X1800XTX selling for $750. The highest I ever seen them was like $650 and I thought that was stupid. I remember some Crazy NV fan-boys paying like $900-$1000 for 7800GTX 512's
Thats the stupidest thing I ever seen fan-boys do.
 
TrueChaos said:
Im in Canada, prices are higher. For some reason ATi cards are a fair bit more than nVidia.

7800GTX -- 589 (price went up a bit o_O)
http://www.infonec.ca/site/main.php?module=detail&id=12417

And the X1800XT -- not crossfire -- 639
http://www.infonec.ca/site/main.php?module=detail&id=13215
Click to expand...
Ooh a fellow Canadian, cool, yeh I was just talking about how biased our X1800 XT, X1900 XT/XTX prices are to the Nvidia equivalents.

eVGA 7900 GT CO Superclock FTW! 550MHZ/1580MHZ Only 399 CAN
http://www.pcvonline.com/productDetails.aspx?id=2479

For the price of that particular X1800 XT the X1900 XTX can actually be had for cheaper!
Though it's still too much, I ain't pay 600 CAN for ATI hardware in additon for me at least Ontario taxes of 15%. :rolleyes:

Sapphire X1900 XTX Standard Clocks 599 CAN
http://www.pcvonline.com/productDetails.aspx?id=2164
 
Fort_Major said:
So what's all this mumble jumble about Conroe :p somebody mentioned about 2.4 "on air" ? Any clue as to what that term means?
Is it really worth waiting till this fall or winter to pick a machine up ? or should one build a system and then ponder upgrading later?
Click to expand...
I am not sure what they meant but Victor Wang was able to run his Conroe 2.4GHZ ES, with a Zalman 9500 without the fan on! So completely passive air cooling. :cool: Conroe runs really cool and doesn't need excessive cooling at all.

I would seriously consider waiting for Conroe it looks to be the most major Intel architecture introduction of the decade, not only that but they will be introduced at good affordable prices.
 
Status
Not open for further replies.
Back
Top