Conroe vs. AMD FX-62

[theprodigalrebel]

Originally posted by: JAG87
I cannot believe idiots like you still compare an architecture that AMD has had for 2 years, to a technology that intel has yet to release.

Sure K8 came out 2 years ago...but that's not the point. The point is that AM2 came out TODAY and is what Core 2 Duo will be competing against. Get the hell over it (I'm not trying to be rude or a jerk...I seriously think you oughta acknowledge that your favorite team didn't win and just let it go)

And for the love of god, don't think about posting something like "Just wait till K8L beats the crap out of Core 2 Duo next year/whenever." For AMD's sake, it better. This round goes to Intel.

 

[rmed64]

Originally posted by: Absolute0
JAG87
Quit complaining, Conroe is here and we're going to compare it to the other stuff.

@ the article
Good stuff. Conroe wins by 40% in one of the gaming tests. The E6600 beat the FX62 in every game test. Beautiful.

Yes, that was the Far Cry test. But they were stupid to put the last 2 games, Splinter Cell and Quake 4 at 1600x1200 res 4xAA and 8xAF/16xAF, because that most likely put the limit on the graphics card, not the cpu, so its not really indicative of total cpu power if card is bottlenecked.

Most likely a Core2duo will be my next chip sometime next year

 

[imported_UncivilizedAMD]

Originally posted by: rmed64

Originally posted by: Absolute0
JAG87
Quit complaining, Conroe is here and we're going to compare it to the other stuff.

@ the article
Good stuff. Conroe wins by 40% in one of the gaming tests. The E6600 beat the FX62 in every game test. Beautiful.

Yes, that was the Far Cry test. But they were stupid to put the last 2 games, Splinter Cell and Quake 4 at 1600x1200 res 4xAA and 8xAF/16xAF, because that most likely put the limit on the graphics card, not the cpu, so its not really indicative of total cpu power if card is bottlenecked.

Most likely a Core2duo will be my next chip sometime next year

But they used the same graphics card for all systems so lets assume that the bottleneck is at the graphics card then all systems would be bottlenecked there. Then the difference in FPS would be attributed to all other components of the system, particularly the CPU and thus it is an indication of total cpu power.

 

[theprodigalrebel]

Originally posted by: rmed64

Yes, that was the Far Cry test. But they were stupid to put the last 2 games, Splinter Cell and Quake 4 at 1600x1200 res 4xAA and 8xAF/16xAF, because that most likely put the limit on the graphics card, not the cpu, so its not really indicative of total cpu power if card is bottlenecked.

You got a point but sometimes, I like to think that if you are interested in high-end CPUs, you are also probably interested in high-end VPUs. All the people looking at AM2/Core 2 are drooling about playing the latest games at 1600x1200 4xAA 16xAF or even higher at the magical min. 60FPS. A 30-40% victory at 1024x768 is a little misleading if all the top games in the market are performing within 3-5% at the res/settings that people ACTUALLY will be playing at.

 

[rmed64]

Originally posted by: theprodigalrebel

Originally posted by: rmed64

Yes, that was the Far Cry test. But they were stupid to put the last 2 games, Splinter Cell and Quake 4 at 1600x1200 res 4xAA and 8xAF/16xAF, because that most likely put the limit on the graphics card, not the cpu, so its not really indicative of total cpu power if card is bottlenecked.

You got a point but sometimes, I like to think that if you are interested in high-end CPUs, you are also probably interested in high-end VPUs. All the people looking at AM2/Core 2 are drooling about playing the latest games at 1600x1200 4xAA 16xAF or even higher at the magical min. 60FPS. A 30-40% victory at 1024x768 is a little misleading if all the top games in the market are performing within 3-5% at the res/settings that people ACTUALLY will be playing at.

I understand your high end cpu : high end card theory, but this is a test about the CPU, not the video card. In truth, each game should have been done as a speed test, just like far cry. Lower settings and less strain on GPU would show which CPU is really faster.

 

[dguy6789]

Name calling is pointless and really shows the maturity level of the posters here. Core 2 Duo will in all liklihood be faster than K8. What is the big deal? Before K8 came out, we could say K8 will in all liklihood be faster than Netburst. Newer technology is generally superior in some way, or else they would have no reason to release it. What I want to see is Core 2 Duo vs K8L.

 

[Nyati13]

Originally posted by: Absolute0
JAG87
Quit complaining, Conroe is here and we're going to compare it to the other stuff.

@ the article
Good stuff. Conroe wins by 40% in one of the gaming tests. The E6600 beat the FX62 in every game test. Beautiful.

No, Conroe isn't here yet. It has not been released for sale, and won't be for several more months.

 

[Lonyo]

Different RAM = stupid article.
Socket AM2 vs Conroe vs Pentium 775's.
All DDR2 based platforms.
Yet they use different RAM (both speed and amount) for the AM2 system.
Lower RAM may hurt Conroe a bit in games, and it would be nice to see either 2 sticks or 4 sticks on both AM2 and Conroe to see the differences, rather than 2 sticks for Conroe and 4 for AMD which doesn't represent a level test bed.

Hexus also shows Conroe having lower memory latency than the AM2 based systems (Sciencemark 2.0)
Xbitlabs show that DDR2 667 @ 3-3-3-10 on an AM2 system has almost the same latency as DDR2-800 @ 5-5-5-10.
The Conroe latency is 3-2-2-8/667 vs 4-3-3-10/800 for AM2 at Hexus.
My idea of logic would suggest that the Core architecture, lacking an onboard memory controller, would have to be slower than an AM2 system.
Anandtech in its comparison of Core and K8 states "With the numbers available to us now, we have reason to believe that the Athlon 64 X2's latency advantage will shrink to only 15 to 20%."
Yet Hexus shows the same or lower latency than AM2 for their Conroe system, which seems odd, though it may be something to do with the specific test.

 

[Lonyo]

Originally posted by: JAG87
you should not compare the two. I understand that there is nothing to compare conroe to, but honestly its like saying ATI X1900 kills the 6800s... yea no ****** sherlock it only came out 2 years later... you just should not compare the two. kudos to conroe and to intel but for the love of god stop bashing the K8 architecture.

AMD are releasing new parts based on that architecture still, and they are aimed at the same price points as the Intel processors.
Comparing an X1900 to a 6800 is a flawed analogy, as the 6800 is at a lower price point and should be compared with something more like the X1600 or lowest end X18/900 cards.

You SHOULD compare K8/AM2 to Conroe because they are the 2 architecures competing with each other.
I don't care if they come from 4 years ago vs 2 months, THEY ARE COMPETING.
And no one is really bashing K8, it's just getting on a bit, same as Netburst was slow and sucked ass when AMD's new K8 architecture was released.

 

[Henny]

Originally posted by: JAG87
I cannot believe idiots like you still compare an architecture that AMD has had for 2 years, to a technology that intel has yet to release.

So AM2 has been out for two years?? I must be behind the times. LOL.

 

[Vee]

There is absolutely nothing wrong with comparing new things to old. On the contrary, it's important to do so. (May I, for instance, suggest you start comparing power consumption of the new AM2 parts to older chips?)

I would interpret those who have raised their voices on this issue as if actually having a completely different point on their minds: The fact that Conroe isn't here, but that AM2 is.

AM2 is no new cpu architecture and as such no direct performance upgrade to 939. That doesn't mean it isn't an improvement. It does come with topend extension to the available performance ratings. And it provides a vastly better performance per Watt. You will now even be able to get a X2 3800+ that ticks along on only a few Watt. DDR2 memory also means that memory power requirements will be lower. Finally there is the virtualization feature.
I don't see anything even remotely competitive available from Intel today.

Core 2 will hopefully be widely available towards the end of this year. But I'm somewhat disappointed in humanity. Due to the effects of Intels marketing trick to showcase Conroe very early. I mean - there's even one poster in this thread who claims: "Conroe is here". Another asks: "Where is AMD response?".

Well, no - Conroe is not here. Not yet. And a "response" cannot precede what it is responding to. Have some pride for C*S*! Don't be so easily manipulated.

Preliminary comparisions of Conroe to current CPUs like the FX62 are of great interest. It helps us understand what potential we can expect from a future technology. But if you interpret these results, as if you're making a customers choice between two available CPUs, then you're a damn fool.

First of all, if you have no need of a computer part today, you're a fool wasting your money on something you don't need. And wasting it on something that by past 30 years experience is going to become obsolete rather quickly.
Further, if you indeed do need a computer part today and have good use for it today, then you're again a damn fool to be waiting. Waiting in order to be a first adopter of a first generation of some new technology.

Let's see now, remember these?:
"Wait for Willamette". "Wait for SATA", "Wait for BTX", "Wait for DDR2", "Wait for Prescott", "Wait for PCIe".

It's not just that waiting for some new technology often will not bring you any tangible additional user value. It's that the real choice is not between FX-62 and any Core 2. The real choice comes the day it's time to buy a new CPU. What will you pay? And for what features and what level of performance? The real choice is the available CPU that fits that slot on that day.

- The "Waiting game" is a game that goes on forever.

 

[SexyK]

Where are people coming up with the idea that Core 2 won't be available until the end of the year??? Everything is pointing to July/August availability.

 

[imported_Questar]

Originally posted by: SexyK
Where are people coming up with the idea that Core 2 won't be available until the end of the year??? Everything is pointing to July/August availability.

That's OEM availability, not retail shelves. I'd bet sometime in the first half of August you'll be able to obtain these.

Some people just need to find any reason to not give Intel credit. They act like Intel succeding is a personal failure for them. Therefore they have to grasp at any straw they can to discredit Intel. So rumors like you can't buy one for six months get started.

 

[Looney]

Originally posted by: JAG87
I cannot believe idiots like you still compare an architecture that AMD has had for 2 years, to a technology that intel has yet to release.

They're comparing what's going to be on the market, so people will know what to buy. Moron.

 

[Henny]

Originally posted by: SexyK
Where are people coming up with the idea that Core 2 won't be available until the end of the year??? Everything is pointing to July/August availability.

Woodcrest = June
Conroe = July
Merom = August

 

[n7]

Hawt numbers.

As Hexus's gaming benches showed, there's little difference at high resolutions/settings, since things are GPU limited.

Therefore, i'll try to hang onto my Opty @ FX60 speeds as long as i can (meaning till i can no longer hold back the upgrade to Core 2 Duo urge)

 

[skooma]

Which will for me be when they hit store shelves

 

[classy]

Hmmm quite a few blunders in this one. First off the FC mb is significantly slower than a comparable Asus board, expecially in gaming. Review numbers of both boards. Still don't know why they use of 4 512 sticks of ram. The bios date in the Hexus FC board is from January. Cliff notes The Conroe adds absolutely nothing in high resolution gaming and in a correct setup may even lose. Then they slip in Fear at a lower resolution with some fantastic numbers no doubt. But how mant people will play Fear at those settings. Some of the other numbers are gaudy as well. But AMD was even losing to the Pentium D 965 in the same benchmarks. At Tech Report they bench a Woodcrest against a lower clocked Opteron setup and it gets by. Cliff notes a higher clocked Opteron probably wins. It seems to me a lot of stuff is over flufffed with Conroe. Its faster, but that 30-40% bs we now know was a lie and add to the fact we still have no date when a person can buy one, no concrete true facts on prices, and no knowledge of the future of the socket. Will we get the same 955 975 915 two and 3 variation nonsense of each requiring a person to buy a new board as well with every processor change? Great numbers with Intel but reality is we all have from Intel is numbers with no product. Can a buy one a month from now, 2 months, will it be Labor day, Thanksgiving, or how about Christmas. I am still using my P4 2.4@2.4 so I am still an Intel guy, but truth is I am so tired of the marketing bs from them, I think I am tired of waiting, its a joke. Intel Just give me something I can buy worthwhile, god its only been 4 freaking years.

 

[deadseasquirrel]

Originally posted by: rmed64

Originally posted by: theprodigalrebel

Originally posted by: rmed64

Yes, that was the Far Cry test. But they were stupid to put the last 2 games, Splinter Cell and Quake 4 at 1600x1200 res 4xAA and 8xAF/16xAF, because that most likely put the limit on the graphics card, not the cpu, so its not really indicative of total cpu power if card is bottlenecked.

You got a point but sometimes, I like to think that if you are interested in high-end CPUs, you are also probably interested in high-end VPUs. All the people looking at AM2/Core 2 are drooling about playing the latest games at 1600x1200 4xAA 16xAF or even higher at the magical min. 60FPS. A 30-40% victory at 1024x768 is a little misleading if all the top games in the market are performing within 3-5% at the res/settings that people ACTUALLY will be playing at.

I understand your high end cpu : high end card theory, but this is a test about the CPU, not the video card. In truth, each game should have been done as a speed test, just like far cry. Lower settings and less strain on GPU would show which CPU is really faster.

True, since they are testing only CPU power, lower res with no AA/AF would be the best way... but my question would then be why test games at all if you're going to test them at settings in which nobody would play them at if they owned such a chip/system? The test then loses any real-world association at that point. IMO, either test games across several resolution settings or don't test them at all.

Me personally-- I use my computer for a lot of different things... http, email, dvd encoding, music, Word, etc. But, primarily, it is used for gaming. And, I really don't care how responsive it is at unzipping a file, or encoding a DVD... but I *do* care how it handles Far Cry at 1600x1200 with AA/AF. These benchmarks (if they hold up for retail) show me that I should simply stick with my existing chip and keep buying the fastest video cards I can... yet remain in a wait-and-see mode. If Oblivion is a sign of the GPU-hungry games of the future, then my CPU will not matter in the least... but, if more games are like Quake4, a CPU/socket upgrade will be ideal.

One thing is for sure... Conroe, Vista, AM2, DX10, G80, R600... there's plenty of good stuff on the way for the inner-geek in all of us.

 

[Vee]

Originally posted by: deadseasquirrel
True, since they are testing only CPU power, lower res with no AA/AF would be the best way... but my question would then be why test games at all if you're going to test them at settings in which nobody would play them at if they owned such a chip/system? The test then loses any real-world association at that point. IMO, either test games across several resolution settings or don't test them at all.

I think the question whether we need Conroe performance to run games comfortably is completely beside the point of both this thread and Conroe's performance. I also think most people here already know the answer to that question.

Benchmarking games is excellent for evaluating CPU performance. That is so because the results are valid and give a true indication of the work-crunching ability of the CPU. Unlike so many other benchmarks. You do of course need to remove the GPU from the results. Running tests at several resolutions doesn't provide any additional information about the CPU.

 

[dexvx]

Originally posted by: Vee

Originally posted by: deadseasquirrel
True, since they are testing only CPU power, lower res with no AA/AF would be the best way... but my question would then be why test games at all if you're going to test them at settings in which nobody would play them at if they owned such a chip/system? The test then loses any real-world association at that point. IMO, either test games across several resolution settings or don't test them at all.

I think the question whether we need Conroe performance to run games comfortably is completely beside the point of both this thread and Conroe's performance. I also think most people here already know the answer to that question.

Benchmarking games is excellent for evaluating CPU performance. That is so because the results are valid and give a true indication of the work-crunching ability of the CPU. Unlike so many other benchmarks. You do of course need to remove the GPU from the results. Running tests at several resolutions doesn't provide any additional information about the CPU.

If that were true, we'd be benching games at 640x480 with the lowest possible settings. But the fact is, its not "real world". If you want to test a CPU's crunching power, other benchmarks are better suited.

 

[ncage]

Koodos to intel for the design of this excellent cpu. I haven't bought an intel cpu in a long time and, with the looks of things, conroe (core 2 duo) will be my first intel cpu in a long time. I love amd but intel will just have the lead for awhile it looks. I currently have an 939 X2 3800+ overclocked to 2.6ghz which i love. I probalby will upgrade to conroe pretty soon after it comes out. I don't do games that much. You ask why do i need this power then? I am a programmer and if any of you have ever stepped through a Visual Studio 2005 program with all the debugging features added you will know its as slow as a dog. At work i have a 3ghz p4 with 2.5ghz of memory and its HORRIBLE. On my system at home its better. Anyways intel is making one mistake here that im sure most people don't care about. They are leaving the current FSB infrastructure intact. They need to come up with something a lot better. Sure its probably not going to matter with 1-2 cpu even if they are dual core but the FSB structure is where AMD has got Intel. The FSB just won't scale. This won't matter to most people but it will matter a lot to the enterpise (and this is where the highest profit margins are). I even think intel could use the hypertransport system if they wanted since amd sent this system through standardized body eventhough im sure they would get crap for following amd rather than leading but who cares just as long as it works. So as i see intel has got what it takes for workstation/light server. But for expensive (high profit margin) servers amd still has intel here. Try to create an 8-way+ on the FSB architechture and watch it fail miserably.

ncage

 

[Accord99]

Originally posted by: ncageBut for expensive (high profit margin) servers amd still has intel here. Try to create an 8-way+ on the FSB architechture and watch it fail miserably.

ncage

It's not easy but IBM and Unisys have, and such systems performs quite well. AMD's Direct Connect architecture doesn't do as well, since they don't have enough links and the HT links aren't fast enough to handle the cache coherency. Opterons can only gluelessly go up to 8S anyways.

 

[Nyati13]

Originally posted by: Lonyo

You SHOULD compare K8/AM2 to Conroe because they are the 2 architecures competing with each other.
I don't care if they come from 4 years ago vs 2 months, THEY ARE COMPETING.

They're not competing yet (and all capping doesn't make you right ). Go try to find a Conroe, it won't be available for months.

When Conroe does show up, it will rule the performance market until K8L shows up, then Intel will figure out something new, then AMD, etc.

 

[Bobthelost]

The reason they used different RAM is that the tests were done at two different labs. I'm far from impressed by the review, one of the most logical benchmarks to run would have been FEAR. That way we could have a common point of reference for comparisons. The tests chosen aren't all that impressive and other than the farcry test the only ones that interested me were the media encoding times, which they did with a 300GB drive vs a 160GB one. Which make them utterly worthless.

There is no comment as to why they are using 4 sticks of RAM for the AM2 rather than 2 sticks. It takes a few hours to drive from the midlands where the tests were run to london, it might have delayed the results by a day because of the inability to test concurrently, (assuming they didn't have one CPU/system before the other) but imo the results would have been far more useful. As it is it's ambiguous.

Edited for excessive and unjustified harshness.