6000+ vs Phenom

[verix]

I've been on the fence lately about upgrading to a X2 6000+, and I'm wondering if anyone has any insight as to whether Phenom will reach that level of performance within, say, 6 months from now?

I ask because I know AMD will be releasing Agena at lower clock speeds initially, but I have no clue as to how quickly they would ramp up. And the other consideration is, of course, price/performance. The 6000+ is at a really nice ratio right now, and I doubt Phenom, being so new, would reach that kind of value anytime soon....but that's why I'm askin, they're supposed to be aggressively priced. And would the launch of Phenom mean some price cuts for the existing X2 line?

Thanks for any of your help!

Verix

 

[GFORCE100]

Originally posted by: verix
I've been on the fence lately about upgrading to a X2 6000+, and I'm wondering if anyone has any insight as to whether Phenom will reach that level of performance within, say, 6 months from now?

I ask because I know AMD will be releasing Agena at lower clock speeds initially, but I have no clue as to how quickly they would ramp up. And the other consideration is, of course, price/performance. The 6000+ is at a really nice ratio right now, and I doubt Phenom, being so new, would reach that kind of value anytime soon....but that's why I'm askin, they're supposed to be aggressively priced. And would the launch of Phenom mean some price cuts for the existing X2 line?

Thanks for any of your help!

Verix

I reckon AMD will be having problems making faster Phenoms/Agenas for some time down to their 65nm production not being up to scratch. The best example of this is that their Athlon X2's above 2.6GHz are all 90nm parts.

AMD needs cash hence prices will be as high as they can be for them to still sell Agens and Phenoms, K10's basically.

If a 2.5GHz K10 requires 1.5V to run, then it's a fine example of how bad 65nm is on such a large die. The fact AMD wants to sell tri-cores is also a result of poor yields of their quad-core K10's. It's basically AMD looking to make the best of a bad situation.

If you must use AMD, for whatever reason then the 6400+ is the best chip right now.

K10's at 1.9/2GHz is just not going to cut it. AMD would need 3GHz models in mass production to be competetive.

 

[SniperDaws]

Go with Intel until AMD get there act together, dont be a fanboy just get the fastest for your money.

 

[Viditor]

Originally posted by: verix
I've been on the fence lately about upgrading to a X2 6000+, and I'm wondering if anyone has any insight as to whether Phenom will reach that level of performance within, say, 6 months from now?

I ask because I know AMD will be releasing Agena at lower clock speeds initially, but I have no clue as to how quickly they would ramp up. And the other consideration is, of course, price/performance. The 6000+ is at a really nice ratio right now, and I doubt Phenom, being so new, would reach that kind of value anytime soon....but that's why I'm askin, they're supposed to be aggressively priced. And would the launch of Phenom mean some price cuts for the existing X2 line?

Thanks for any of your help!

Verix

I really don't think anyone knows this yet...not even AMD completely.
The Agena was already demoed at 3 GHz in June, though whether they release at that speed is still under question. I think it will depend on the latest stepping (B3) and how well it does.

 

[darkfalz]

Don't buy AMD full stop, their chips are garbage.

 

[bryanW1995]

ok, based upon what verix has written, it is reasonable to assume that he has an am2 system right now and just wants to drop in another cpu. that precludes intel and all the hassles involved with changing an entire system. So, the question becomes, do you go 6000+, 6400+, or wait for phenom?

Look at it this way, guys. How many of you would prefer an e6850 to a Q6600? Not many, right? Phenom is supposed to be at least 15% better clock for clock than x2, so for comparison's sake, a 2.1 phenom vs 6000+ would, comparatively, be like a Q6600 vs e6850. You should be able to get a 2.1 phenom for about the same price in 3-4 mos that you could get a 6000+ right now. The phenom will probably produce less heat, have 2 more cores, and in general be much more future-proof. If you can wait the 3-4 months then do it. If you must upgrade now then get the 6000+. The 6400+ looks nice, but it's about 50 % more expensive for 7% more clock speed.

 

[Viditor]

Originally posted by: bryanW1995
ok, based upon what verix has written, it is reasonable to assume that he has an am2 system right now and just wants to drop in another cpu. that precludes intel and all the hassles involved with changing an entire system. So, the question becomes, do you go 6000+, 6400+, or wait for phenom?

Look at it this way, guys. How many of you would prefer an e6850 to a Q6600? Not many, right? Phenom is supposed to be at least 15% better clock for clock than x2, so for comparison's sake, a 2.1 phenom vs 6000+ would, comparatively, be like a Q6600 vs e6850. You should be able to get a 2.1 phenom for about the same price in 3-4 mos that you could get a 6000+ right now. The phenom will probably produce less heat, have 2 more cores, and in general be much more future-proof. If you can wait the 3-4 months then do it. If you must upgrade now then get the 6000+. The 6400+ looks nice, but it's about 50 % more expensive for 7% more clock speed.

QFT
Excellent summation bryan...

 

[Keysplayr]

Originally posted by: darkfalz
Don't buy AMD full stop, their chips are garbage.

No. No they aren't. Just not as good as C2D as P4/PD was not as good as A64/X2.
So, AMD chips are FAAAAAAR from garbage. In some ways, they are more advanced than C2D, and the upcoming Penryn.

 

[verix]

Thanks everyone for you help and quick replies! And thank you, Bryan, for getting to the heart of what I was after. I'm still kinda playing around with it in my head of whether I want instant gratification or whether I wanna wait and perhaps upgrade my motherboard too with an Intel combo. Your comments will definitely help in weighing pros and cons!

 

[bradley]

Originally posted by: keysplayr2003

Originally posted by: darkfalz
Don't buy AMD full stop, their chips are garbage.

No. No they aren't. Just not as good as C2D as P4/PD was not as good as A64/X2.
So, AMD chips are FAAAAAAR from garbage. In some ways, they are more advanced than C2D, and the upcoming Penryn.

Let's even the argument out a little bit. If AMD too had stitched multiple dies onto the same substrate maybe we wouldn't even be having this conversation. But we live in a world where everyone wants quantity over quality, speed over almost every other factor; and they want it now.

Certainly having a more elegant native dual and quad-core design, and an on-die memory controller, theoretically makes the X2 and Phenom far more advanced on paper, a feat currently unmatched by Intel. These two factors alone mean less computational overhead, less heat, and less landfill waste. Just consider today how every Intel upgrade will also require a new motherboard, and updated northbridge. No thanks. That's just my opinion though.

 

[zach0624]

Originally posted by: Viditor

Originally posted by: bryanW1995
ok, based upon what verix has written, it is reasonable to assume that he has an am2 system right now and just wants to drop in another cpu. that precludes intel and all the hassles involved with changing an entire system. So, the question becomes, do you go 6000+, 6400+, or wait for phenom?

Look at it this way, guys. How many of you would prefer an e6850 to a Q6600? Not many, right? Phenom is supposed to be at least 15% better clock for clock than x2, so for comparison's sake, a 2.1 phenom vs 6000+ would, comparatively, be like a Q6600 vs e6850. You should be able to get a 2.1 phenom for about the same price in 3-4 mos that you could get a 6000+ right now. The phenom will probably produce less heat, have 2 more cores, and in general be much more future-proof. If you can wait the 3-4 months then do it. If you must upgrade now then get the 6000+. The 6400+ looks nice, but it's about 50 % more expensive for 7% more clock speed.

QFT
Excellent summation bryan...

Agreed bryan pretty much summed it up right there I would go for the phenom (why I upgraded to am2) which should be alot cooler than a 6000+ or 6400+(The barcelona review hints at this plus it will be a new stepping) plus 2 more cores so it looks to be a better deal than a 6000+ is now and when phenom comes out there shouldn't be a reason to get a 90nm x2.

 

[bryanW1995]

verix, what do you use your system for? how long do you intend to keep it? do you plan to make any changes in your usage pattern during its expected lifetime?

 

[Keysplayr]

Originally posted by: bradley

Originally posted by: keysplayr2003

Originally posted by: darkfalz
Don't buy AMD full stop, their chips are garbage.

No. No they aren't. Just not as good as C2D as P4/PD was not as good as A64/X2.
So, AMD chips are FAAAAAAR from garbage. In some ways, they are more advanced than C2D, and the upcoming Penryn.

Let's even the argument out a little bit. If AMD too had stitched multiple dies onto the same substrate maybe we wouldn't even be having this conversation. But we live in a world where everyone wants quantity over quality, speed over almost every other factor; and they want it now.

Certainly having a more elegant native dual and quad-core design, and an on-die memory controller, theoretically makes the X2 and Phenom far more advanced on paper, a feat currently unmatched by Intel. These two factors alone mean less computational overhead, less heat, and less landfill waste. Just consider today how every Intel upgrade will also require a new motherboard, and updated northbridge. No thanks. That's just my opinion though.

No, that argument of "elegance" is non-existant. If the performance is there, it does not matter if two cores are "glued" together, or on the same piece of silicon. Performance talks and BS walks. As far as the "more advanced" comment I made, I was referring only to the integrated memory controller AMD employs. Very nicely done. But, this does not help against C2D architecture and extra cache.

 

[Markfw]

keys, the "advanced" AMD architecture with IMC and HT does to this day beat Intel in 4x and greater cpu configs in servers.

Now for desktop, Intel is clearly leading until AMD can ramp speeds up. And I have 4 Intel quads, so I am not AMD biased, but aside from clockspeed, they do have the better design....IMO

 

[Arkaign]

Originally posted by: Markfw900
keys, the "advanced" AMD architecture with IMC and HT does to this day beat Intel in 4x and greater cpu configs in servers.

Now for desktop, Intel is clearly leading until AMD can ramp speeds up. And I have 4 Intel quads, so I am not AMD biased, but aside from clockspeed, they do have the better design....IMO

Well, to be fair, judging from pure results :

AMD's design is better for servers
Intel's design is better for desktops

Am I missing something? The design means nothing without results. If Plutonium processors made a 50ghz 486 that was faster than a 5Ghz C2Q, then it would be a better design, regardless of "elegance"/etc.

 

[Viditor]

Originally posted by: Markfw900
keys, the "advanced" AMD architecture with IMC and HT does to this day beat Intel in 4x and greater cpu configs in servers.

Now for desktop, Intel is clearly leading until AMD can ramp speeds up. And I have 4 Intel quads, so I am not AMD biased, but aside from clockspeed, they do have the better design....IMO

Personally, I agree with both of you...
I agree with keys that at the end of the day, you buy the processer/platform that has the best price/performance for the job at hand.
I also agree with Mark that because of the architecture, often this is AMD...but that architecture doesn't really come into play much at under 8 cores, so desktop performance (though not necessarily price/performance) is still solidly in Intel's corner.

I am convinced that AMD's lead in servers will extend much further with Barcelona until Nehalem is released (at that point it's a big question mark) due to the architecture...I don't think Tigerton is going to be very effective at all.
The desktop though is very murky at the moment, mainly because none of us has any idea what's actually going to be released in November yet (I very much feel Anand's pain on this.) In fact, we have yet to see a review using the actual shipping silicon for Barcelona (stepping BA).

However it breaks down, this certainly is a very interesting year!

 

[bryanW1995]

@arkaign, actually, 50 ghz 486's would probably need plutonium reactors to power them, so they would not be a better design. we are more and more these days seeing people look at power consumption/performance/heat as a whole rather than just performance. I still remember how shocked I was at the incredible room temp increase when I went from an fx55 to an opty 180, for example.

also, I think that you just re-stated exactly what mark said. 4x + servers are still ruled by amd, desktops are still ruled by intel. That has no chance of changing until nehalem comes out, and who knows when intel will roll that out in 4x + servers.

 

[Keysplayr]

Originally posted by: Markfw900
keys, the "advanced" AMD architecture with IMC and HT does to this day beat Intel in 4x and greater cpu configs in servers.

Now for desktop, Intel is clearly leading until AMD can ramp speeds up. And I have 4 Intel quads, so I am not AMD biased, but aside from clockspeed, they do have the better design....IMO

Agreed Mark. I was specifically talking desktops.

 

[Idontcare]

Originally posted by: Arkaign

Originally posted by: Markfw900
keys, the "advanced" AMD architecture with IMC and HT does to this day beat Intel in 4x and greater cpu configs in servers.

Now for desktop, Intel is clearly leading until AMD can ramp speeds up. And I have 4 Intel quads, so I am not AMD biased, but aside from clockspeed, they do have the better design....IMO

Well, to be fair, judging from pure results :

AMD's design is better for servers
Intel's design is better for desktops

Am I missing something? The design means nothing without results. If Plutonium processors made a 50ghz 486 that was faster than a 5Ghz C2Q, then it would be a better design, regardless of "elegance"/etc.

Unfortunately for AMD the server market does not appear to pay the bills as it were.

Also quite the ironic example you gave there...given that AMD ramped their 486 processor line to 133MHz and gave them a "performance rating" of a Pentium at 75MHz.

At that time given the choice between a Pentium 66MHz or an AMD 486 P75 I suppose the customer was enticed to go 486.

 

[bradley]

I'm just saying the overall market is far less cut-and-dried than some enthusiasts are trying to portray. All of those extra cores, various caches, processes, protocols, specifications, memory controllers, chipsets etc. have complicated the benefits greatly in the past several years. Every decision also has its share of trade-offs.

Native cores will be more efficient in communicating with one another. More importantly there are far less diminishing returns with a native design as you add additional cores... in theory it's probably quad-core as the sticking point, but who knows how differently it will execute under real-world usage. Certainly Intel knows the deal with the native design Nehalem. I'm not sure how much longer L3 cache could have fixed things.

I totally understand that a larger die means a higher rate of failure and cuts into AMD's profits. Which is why tri-core is probably almost a necessity for AMD. What I can't understand is how native multi-core design doesn't make AMD's processors more advanced or sophisticated, especially in the server market.

 

[bryanW1995]

amd is more advanced in many ways. most of those things help them to perform better in the server market. intel is more advanced in many ways, too. most of their strengths help them to dominate the desktop/mobile market.

 

[coldpower27]

Originally posted by: bradley
I totally understand that a larger die means a higher rate of failure and cuts into AMD's profits. Which is why tri-core is probably almost a necessity for AMD. What I can't understand is how native multi-core design doesn't make AMD's processors more advanced or sophisticated, especially in the server market.

Simple, the people here haven't set a universally agreed upon definition on what constitutes "advanced", "elegant", "sophisticated".

To me the better performing processor is the one that is more advanced (regardless of how it get's there), and Intel for now wins that ball park even in 4P+ Servers with the Tigerton X7350, until AMD releases their higher clocked Barcelona's.

Elegance is irrelevant if the performance isn't there, but that is my perspective on the issue.

If you can argue that AMD has "more advanced" technology in the form of it's IMC & HT, then Intel has "more advanced" cache technology (larger and faster) and memory disambiguation.

 

[bradley]

AMD made the decision to have native core be a focal point on their limited budget. They don't have the funds and personnel of an Intel to take on as many offshoot projects. This technological advancement has played a HUGE role in why AMD is perceived as trailing Intel, and is totally germane.

Intel probably saw AMD's roadmap and decided to hit them where they lived. Those Intel price cuts you're all enjoying are intended to cripple AMD. Right now native core processors are seen as a huge mistake by AMD -- that and their acquisition of ATI. Obviously lots of people have forgotten already about the mistake that was Pentium D.

AMD's decision made them beholden to the server market, where there is lots of money to be had. Perhaps Barcelona gives AMD further entrance into the server market pre-Nehalem. This is where the Dept. of Justice will play a huge part. Tri-core might fill another void in the desktop market. And AMD still has resellers like Dell and Toshiba.

I just find it hard to believe anyone who talks in definitive terms about silicon that either only exists on paper, or has been demonstrated months away from a launch. And if you give the benefit of a doubt to Intel, than it should also go to AMD. Though the original question was about whether it was worth waiting for Phenom, not at all about Intel. If we're being intellectually honest, right now it's a draw for me... too many unknown variables.