According to Tom's Hardware FX is a failure

[Cleeve]

Hey guys,

I'm Don Woligroski, and I wrote the article.

For the record, i wrote it of my own volition. I have nobody breathing down my neck, promising hardware or ad dollars. Tom's is big enough that we don't have to deal with that bull$hit.

If you want to believe I'm anti-AMD, whatever floats your boat. But you should know that I'm also the guy who writes the monthly "best graphics card for the money" article (completely dominated by AMD Radeons right now), and I also write the monthly "best gaming CPU for the money" article (which, until Sandy Bridge, was dominated by AMD in the sub-$200 range).

I simply think the FX is inferior to Sandy Bridge when it comes to price/performance, and while it does OK-ish in some disciplines it's completely humiliated in others. You can disagree with me of course, but for what it's worth that's my rationale, plain and simple. Just thought I'd let you knwo where I'm coming from.

I root for AMD, honestly I do. I'm hoping that they fix Piledriver, like they fixed the Phenom II compared to the original Phenom. Unfortunately, they can't throw cache at it this time around because there's so much on die already.

Best regards,

- Don Woligroski

[abinstein]

I'm interested to learn from you exactly where Bulldozer is "humiliated" by Sandy Bridge?
I truly hope to find an example which you give to justify that "failure" descriptor.
Not running Cinebench or 3Dmark with as high scores does not.

[Cleeve]

I'm interested to learn from you exactly where Bulldozer is "humiliated" by Sandy Bridge?
I truly hope to find an example which you give to justify that "failure" descriptor.
Not running Cinebench or 3Dmark with as high scores does not.

Running games is one real-world discipline where the FX fails - a SB-based $80 Pentium G630 will beat a $200 FX-8120 for less than half the price.
http://www.tomshardware.com/reviews/gam ... 20-10.html

There are some compression and encoding apps the FX doesn't handle well, although granted there are some it handles beautifully. Overclocking delivers good MHz, but the Sandy Bridge has a lot more legs when overclocked, not raw MHz but when it comes to IPC it's impressive.

For me it's price/performance/power usage in general. You certainly have the right to disagree, I've never suggested you have to see it my way. But I'm calling it as I see it.

[abinstein]

I'm interested to learn from you exactly where Bulldozer is "humiliated" by Sandy Bridge?
I truly hope to find an example which you give to justify that "failure" descriptor.
Not running Cinebench or 3Dmark with as high scores does not.

Running games is one real-world discipline where the FX fails - a SB-based $80 Pentium G630 will beat a $200 FX-8120 for less than half the price.
http://www.tomshardware.com/reviews/gam ... 20-10.html

Your comparison there is completely worthless to the evaluation of FX in general, because--
(1) it is only for gaming, that is just one application that a CPU is made to run. not running games as well in some particularly selected scenarios does not make a processor "failure."
(2) it shows an average of several games with extremely different characteristics. the selection and weight in the types of games can great affect the average.
(3) most of those framerates are not taken from steady-state gameplay. In particular, the StarCraft II framerates are ridiculous. the numbers, as a result, do not convey actual gaming experience.

In any rate, none of those shows FX being a failure. Of course, if someone would spend $550 on such graphics card, he would probably also spend $80 extra for FX-8120p rather than Pentium G630 for better performance on non-gaming applications. Even if he only plays games, I'm sure there are plenty of other games where FX-8120p out-performs Pentium G630.

There are some compression and encoding apps the FX doesn't handle well, although granted there are some it handles beautifully. Overclocking delivers good MHz, but the Sandy Bridge has a lot more legs when overclocked, not raw MHz but when it comes to IPC it's impressive.

So, FX and SB trade punches on compression and encoding. That doesn't make FX more a failure than SB, right?

And there is that "IPC" argument again. Exactly what application you see FX's IPC being a "failure"? I reckon that on average, an FX core has lower IPC than an SB core, and perhaps roughly equal IPC to K10 depending on the workload. Does that make FX a failure, when it gives you 2x more cores per chip?

For me it's price/performance/power usage in general. You certainly have the right to disagree, I've never suggested you have to see it my way. But I'm calling it as I see it.

For me it's biased choice application + biased rationale + biased evaluation. The more serious failure seems to be how you drew the "failure" conclusion by such weak support (and believed it to be "rational" yourself).

[eaima]

Hey guys,

I'm Don Woligroski, and I wrote the article.

For the record, i wrote it of my own volition. I have nobody breathing down my neck, promising hardware or ad dollars. Tom's is big enough that we don't have to deal with that bull$hit.

If you want to believe I'm anti-AMD, whatever floats your boat. But you should know that I'm also the guy who writes the monthly "best graphics card for the money" article (completely dominated by AMD Radeons right now), and I also write the monthly "best gaming CPU for the money" article (which, until Sandy Bridge, was dominated by AMD in the sub-$200 range).

I simply think the FX is inferior to Sandy Bridge when it comes to price/performance, and while it does OK-ish in some disciplines it's completely humiliated in others. You can disagree with me of course, but for what it's worth that's my rationale, plain and simple. Just thought I'd let you knwo where I'm coming from.

I root for AMD, honestly I do. I'm hoping that they fix Piledriver, like they fixed the Phenom II compared to the original Phenom. Unfortunately, they can't throw cache at it this time around because there's so much on die already.

Best regards,

- Don Woligroski

Hi Cleeve, welcome to the zone. I am very happy that you decide to discuss about your article. It was just very funny to read, except to the choice of putting fx cpu here. (although the phenom bug is more about a bug being poorly handled than a technical failure; I must be too used to see it everywhere...)

Architecture and software optimization and performance are very complex, would you call a processor a technical failure because the software you used as a mesurement can't use all the features of the processor? Or the algorithm not being optimized for it?

Besides all of this, I simply disagree because as I said in an earlier post it's too early to call it a technical failure. Lets take the intel atom. IMHO I would call it a technical failure not because bobcat washed it away in term of price/performance, but because it technically failed to conquer the mobile market. With all the money invested, it just failed.

Fx in contrast have not meet the initial goal fixed, with a 2+years delay to market. When amd announced that it was re-introducing its fx branding, me and a lot of people -it seems- were expecting a different desktop performance (I am not gaming btw). I got an emotive reaction to this and in my opinion the marketing department made an horrible job for it. And while I understand that you take the products as what is currently shipped, keep in mind that globalfoundries 32nm is not fully mature yet while the intel one is well oiled by now, so some characteristics are not necessarily what they should be. If fx were on the market for a while then whether any what you would consider as a "lack of competivity" would came from the architecture or fab process would be irelavent, an resonable period of time would have passed. Although adding the word "for gaming" (by supposing that it is where it would fail) would be very appropriate. Lets say something like: 'The fx for gaming: a technical failure'.(And BD arch is not a uarch designed for gaming btw)

As such, I would just agree to call fx the deception of the year! And I would put it on the top of the list. (but it's not a technical failure for me, unlike all the others products on your list, this one is all young and it have improvement coming. Lets see after they shift away from bulldozer.)

[Cleeve]

Your comparison there is completely worthless to the evaluation of FX in general

I disagree with your reasoning. We'll have to agree to disagree I guess.

[Cleeve]

Hi Cleeve, welcome to the zone. I am very happy that you decide to discuss about your article. It was just very funny to read, except to the choice of putting fx cpu here. (although the phenom bug is more about a bug being poorly handled than a technical failure; I must be too used to see it everywhere...)

Architecture and software optimization and performance are very complex, would you call a processor a technical failure because the software you used as a mesurement can't use all the features of the processor? Or the algorithm not being optimized for it?

Besides all of this, I simply disagree because as I said in an earlier post it's too early to call it a technical failure. Lets take the intel atom. IMHO I would call it a technical failure not because bobcat washed it away in term of price/performance, but because it technically failed to conquer the mobile market. With all the money invested, it just failed.

Hey man, thanks for the welcome!

I hear where you're coming from, and I don't disagree with anything in particular. But I do think it comes down to semantics... technical failure vs. epic fail, with the second being a little more tongue in cheek than the first.

Frankly, the article doesn't take itself as seriously I think some of you gents might be taking the FX commentary in there. I mean, the iSmell and Logitech keyboard are listed in the same article.

I do understand this is an AMDZone forum and as such you guys might be a little more sensitive to what you perceive as an attack on AMD. But I thought I'd drop by and give you my two cents anyway, for what it's worth. But if you think I'm anti-AMD, take a look at my best graphics cards values article. As you can see, it's simply not the case:

http://www.tomshardware.com/reviews/gam ... ,3107.html

Kind regards,

- Don

[DamnYank]

I'm interested to learn from you exactly where Bulldozer is "humiliated" by Sandy Bridge?
I truly hope to find an example which you give to justify that "failure" descriptor.
Not running Cinebench or 3Dmark with as high scores does not.

Running games is one real-world discipline where the FX fails - a SB-based $80 Pentium G630 will beat a $200 FX-8120 for less than half the price.
http://www.tomshardware.com/reviews/gam ... 20-10.html

There are some compression and encoding apps the FX doesn't handle well, although granted there are some it handles beautifully. Overclocking delivers good MHz, but the Sandy Bridge has a lot more legs when overclocked, not raw MHz but when it comes to IPC it's impressive.

For me it's price/performance/power usage in general. You certainly have the right to disagree, I've never suggested you have to see it my way. But I'm calling it as I see it.

Cray sees it differently.... strange how relative performance changes when using code that is vendor neutral rather than optimized for the Intel chip...

[wuttz]

Best regards,

- Don Woligroski

welcome to the zone toms.
please have your defibrillator at the ready when you log-in.
why was the 2500k overclocked to 4GHz?
why wasnt any AMD CPU overclocked to 4GHz instead of the 2500K?

intel dont like AMD CPU at top of bar chart?
will it breach marketing contract?

[GummiRaccoon]

I don't think it is a bad list overall, however, I would have included windows ME. And I think people feel the bulldozer is a failure because it is fresh in their minds, to me it was a little bit disappointing that they weren't clocked higher and used less power, I was going to get one but everywhere was sold out when I was upgrading so I got a phenom II 955.

From a gamer's perspective I'd say the 5800 FX was probably the biggest flop of all time, with the 2900XT being close. And 3DFX closing their doors was a pretty big failure in my book.

Also welcome to the zone.

[Cleeve]

why was the 2500k overclocked to 4GHz?
why wasnt any AMD CPU overclocked to 4GHz instead of the 2500K?

intel dont like AMD CPU at top of bar chart?
will it breach marketing contract?

Let's be frank, all Intel needed was the i3-2100 to be 'at the top of the bar chart' when it comes to stock gaming CPUs, so that's not a valid reason to include the OC'd 2500K, is it?

The 2500K was OC'd to 4 GHz to see if there is a benefit to paying more and overclocking a bit. Actually there wasn't, as the $180 i5-2400 almost kept up at stock clocks. That's my recommendation for a gaming CPU at this time.

As far as what Intel thinks, though, I couldn't tell you, they don't do much in the way of PR or talking to editors. Frankly, I think AMD spends a lot more for advertising on our site from what I've seen in our ads, but I couldn't say for sure as the writers are disconnected from the ad department.

If you're suggesting the only way I would recommend an Intel CPU over AMD is by being paid off, I don't think the data supports your theory. But whatever floats your boat.

[Cleeve]

I don't think it is a bad list overall, however, I would have included windows ME.

I almost did. I actually wrote it up in the draft, but it's been so long and I dumped it so quickly to go back to 98 SE back in the day that I couldn't reliably recall the details of why it sucked so bad, so i thought I'd skip that one.

[wuttz]

But whatever floats your boat.

right..


or maybe the 2500k data just wasnt much of an improvement over the 2400.
and the performance difference across games were practically negligible.
so when did you decide to use the 2500K-OC data instead of the 2500K-stocks?
curious w/ the thought process, really.
why not choose the quad-core phenom OC'd to 4+GHz instead?
and the 2500k at stock-
you can do it that way too you know..
and it would be different than all the other same ol same ol reviews everywhere.

[Cleeve]

or maybe the 2500k data just wasnt much of an improvement over the 2400.
and the performance difference across games were practically negligible.

Um, well, yeah... that's kind of what I just said, isn't it? Or are you just agreeing with me?

so when did you decide to use the 2500K-OC data instead of the 2500K-stocks?
curious w/ the thought process, really.

I'm happy to share the thought process (although i thought I already did in a previous post, maybe you missed it?).

I'll explain again tho, looks like you'd appreciate it if I broke it down for you: I was hoping to find out if there's a benefit to paying more than $200 for an unlocked 2500K and overclocking a bit. Looking for the cost/benefit ratio there. In the old days the CPU didn't matter as much as the graphics card, so I wanted to see if there was a point to paying more and pushing the hardware a little.

The results suggested there's not much point over a sub-$200 stock i5-2400. I thought that was cool to know. If access to extra data offends you I don't know what to tell you... I guess we look at it differently. I tend to run tests for the sake of curiosity, not to make someone's hardware look better or worse.

Think about it for a second: if I was out to make AMD look bad I'd have stopped at the i3-2100, that wouldn't have been flattering at all.

why not choose the quad-core phenom OC'd to 4+GHz instead?
and the 2500k at stock-
you can do it that way too you know..
and it would be different than all the other same ol same ol reviews everywhere.

Um... did you read the review? Because, I *did* overclock the 955 to 4 GHz.

The order of operations is: read it through first, then criticize.

[wuttz]

Um... did you read the review? Because, I *did* overclock the 955 to 4 GHz.

The order of operations is: read it through first, then criticize.

read again don,

why not choose the quad-core phenom OC'd to 4+GHz instead?
and the 2500k at stock-

sorry if not clear enough, but this is where i was pointing to;



an AMD CPU w/ a 4GHz OC would look nice there too, right?

[Cleeve]

read again don

read again, wuttz, this is what i was pointing to:

http://www.tomshardware.com/reviews/gam ... 120-9.html

There's no hiding what a 4 GHz Phenom II can do, it's there.

The i5-2500K is the 'control' used throughout the benches to show, as I said, comparative performance to the top-of-the-line. Even when oc'd to 4 GHz, the 955 barely met the stock i3-2100.

Why would I use a 4 GHz 955 to demonstrate the highest performance mark throughout the gamut of tests when it barely reaches Intel's stock $120 CPU at stock clocks? And when the 955 is being end-of-lifed?

That doesn't make sense, bro.

Plus, the 980 is in there. That extra 300 MHz doesn't perform any miracles, but you know that.

I get the feeling the only way you'd be happy with this review is if I overclocked the Phenom II and threw out any data suggesting that a comparably-priced Intel CPU was faster. Which one of us has an agenda to push, again?

[wuttz]

The i5-2500K is the 'control' used throughout the benches to show, as I said, comparative performance to the top-of-the-line. Even when oc'd to 4 GHz, the 955 barely met the stock i3-2100.

rofl!!
i think the only thing in control here is the reader's "perception."
i think the glaring omission in there is the non-mention of the 2500k, really.
practically negligible performance improvement, and even stinks price/perf in bf3 OC'd.

how about we focus a bit on that perception?

[Cleeve]

i think the only thing in control here is the reader's "perception."

Well I guess that's one way of looking at it.

From your point of view, I'm an Intel shill who just influenced upwards of a million viewers toward the great satan on CPUs. From my point of view, you're just a guy who stands behind a brand flag who'd prefer to keep unflattering data from the public.

I suspect you're the self delusional one, but like I said, whatever floats your boat. In any case, I'm pleased you're unable to prevent me from sharing as much data as possible to as many folks as I can reach from one of the most read tech sites on the 'net.

[AussieFX]

I'm interested to learn from you exactly where Bulldozer is "humiliated" by Sandy Bridge?
I truly hope to find an example which you give to justify that "failure" descriptor.
Not running Cinebench or 3Dmark with as high scores does not.

Running games is one real-world discipline where the FX fails - a SB-based $80 Pentium G630 will beat a $200 FX-8120 for less than half the price.
http://www.tomshardware.com/reviews/gam ... 20-10.html

By your standards then anything above a Pentium is pointless/worthless?

There are some compression and encoding apps the FX doesn't handle well, although granted there are some it handles beautifully. Overclocking delivers good MHz, but the Sandy Bridge has a lot more legs when overclocked, not raw MHz but when it comes to IPC it's impressive.

There are compression and encoding programs that intel don't handle well too - no mention of them.
You fail to show/explain where BD does well in encoding compression.
Are you saying BD doesn't OC well?
Are you saying SB scales better with frequency than BD?

For me it's price/performance/power usage in general. You certainly have the right to disagree, I've never suggested you have to see it my way. But I'm calling it as I see it.

About time you revisited your purchases then. My 8120 was $120 less than a 2500K yet performs better. Go figure.
Power consumption, yes BD is currently too high.

[AussieFX]

Your comparison there is completely worthless to the evaluation of FX in general

I disagree with your reasoning. We'll have to agree to disagree I guess.

Why do you disagree?

You've bashed a piece of hardware and you can't explain why except for spurious and blatantly false claims about encoding performance that you hurriedly backtracked on when confronted.
I won't even comment on that gaming farce.

We would like to hear your sound scientific reasons Don.

Or are you just making up stories?

------------------------------------------

Little wonder that tech writing is in the current poor state it is.
Doesn't give me much hope for the future.

...But whatever floats your boat right.

[Cleeve]

By your standards then anything above a Pentium is pointless/worthless?

No. For gaming at least I'd reserve that distinction for an i5-2400, if you wanted to go higher-end with an overclocked i5-2500K you'd probably have to be running CrossFire/SLI to justify it.

About time you revisited your purchases then.

Not much to revisit, as all my hardware is free. I'm running the newest AMD and Intel CPUs as both companies provide processor samples for my testing pleasure.

I won't even comment on that gaming farce.

We would like to hear your sound scientific reasons Don.

You ask for scientific reasons and but simply discount a valid gaming comparison out of hand in the same breath.

Frankly, that's why it'd be unwise to waste time on you. It seems you're looking for a fight more than an actual dialogue. I've stated my reasons to my satisfaction and if you don't like them that's something you'll have to deal with I suppose. Unpleasant for you perhaps, but that's your reality.

[AussieFX]

You ask for scientific reasons and then simply discount a valid gaming comparison out of hand.

Valid gaming comparison...
don't make me laugh.

Frankly, that's why I'm not going to waste time on you. I've stated my reasons to my satisfaction and if you don't like them that's something you'll have to deal with I suppose. Unpleasant for you pernhaps, but that's your reality.

Just what I expected from you. No answer.
Funny that's the same tactic all shills/fuders use.

Nothing is unpleasant for me, I simply know what a fool you've made of yourself.
You say BD is one of the great tech failures but have absolutely NOTHING to back it up.
Expect to be ridiculed, this isn't some kiddy site where everything written must be gospel.

[Cleeve]

Valid gaming comparison...
don't make me laugh.

...

No answer.
Funny that's the same tactic all shills/fuders use.

It certainly is, which is why I'm not surprised that you don't provide a valid technical complaint about the gaming comparison.

Good night sir, I'm sure you'll have some angry verbal garbage for me to refute tomorrow. Have a good time with that.

[AussieFX]

http://www.tweaktown.com/reviews/4350/a ... ndex3.html
http://www.tweaktown.com/reviews/4350/a ... ndex4.html
http://www.tweaktown.com/reviews/4350/a ... ndex5.html
http://www.tweaktown.com/reviews/4350/a ... ndex6.html

That seems to fly in the face of your gaming graph.

I'm really curious to know your reasoning for placing it in the top 10 failures. There are other AMD designs I'd throw in before BD.

[gallier2]

By including only 1 OC result, without comparison with its stock result, the whole rationale crumbles.
And for the power consumption of FX, don't forget that Intel and AMD do not count the same. On Intel the wattage is only for the core, the uncore part is not included, on AMD the wattage is for the whole socket.
http://www.lostcircuits.com/mambo//inde ... 4&Itemid=1
viewtopic.php?f=532&t=138958&p=214637#p214617