Impressed with FX-8350 and the new article at Anand

[guskline]

Sure I too am fortunate to have an MC and Frys local to me, so yes at $229 the 3770K does look good. However you get $40off a mobo if you buy the AMD chip .

Looks like I'll be building a new rig with the FX for my bro. He has an aging Amd X4 820 I think.

Seriously consider the I5-3570k bundle. I had replaced my 2500k with the 3770k in rig 1 below. I also had the 2500k OC'd to 4.4Ghz. It was slightly faster than the 8350. I would consider them about equal for gaming. Obviously the 3770k widens the performance gap.

The other thing to consider is the 8350/8320 really need the mbs with better phase control to have successful OCs as they really draw power when OC'd. I would suggest the higher priced mbs. The Asus Sabertooth is the most expensive but I use that mb for BOTH my 8350 and my 8320 and it is built like a tank.

The 3570k is not as sensitive to the quality of mb for OCs ( it needs a better heatsink!). You could actually buy a less expensive 3570k bundle and likely get better gaming performance.

I notice from your sig you are well aquainted with high end computer parts so you will have the skill and knowledge to make the correct choice for you.

Good luck with the build and keep us posted with your results.

 

[Chiropteran]

If you move the bottleneck, you can make almost any CPU look good.

Move? Isn't it pretty much a given that the average gamer is going to run around with a single GPU system? Silly SLI $3000 setups are the exception, not the rule.

The bottleneck is naturally the GPU, it is only moved to the CPU in fake artificial low res tests that don't actually apply to the real world.

 

[BrightCandle]

I have two issues with this review.
1) Is a big one - average FPS with a single number. No attempt to look at even minimums or the range of frame rates. Tomshardware did a significantly better review some months ago testing the 8350 and showed while averages were similar there were problems with consistent frame delivery, the top and bottom frame rate delivered was different. Doing a single FPS test in today's tooling is unacceptable because we know the data is often misleading.

2) The range of games used was hardly a broad range of different types of games. In essence what we had was 3 known GPU limited games with very light CPU usage and 1 heavy CPU user. It isn't a sufficient list of games to say much other than that this very odd selection works the way it does.

I have plenty of games that are clearly CPU limited at 4.5Ghz and 2 680's. It is a real struggle to get to 120 fps, and that is often dominated by the CPU not the GPUs.

 

[Ventanni]

I'm impressed with the FX-8350 as a gaming chip overall, but I'm not impressed with it in these benchmarks for the simple reason that if you're going to spend the money on a multi-GPU setup, you'd be dumb not to spend the extra money on the fastest CPU available to fuel them.

I do feel though that the power consumption difference between the 8350 and the 3770K is much less of a factor in a multi-GPU setup considering it's the GPU's that are using the majority of the system power.

 

[UNhooked]

Seriously consider the I5-3570k bundle. I had replaced my 2500k with the 3770k in rig 1 below. I also had the 2500k OC'd to 4.4Ghz. It was slightly faster than the 8350. I would consider them about equal for gaming. Obviously the 3770k widens the performance gap.

The other thing to consider is the 8350/8320 really need the mbs with better phase control to have successful OCs as they really draw power when OC'd. I would suggest the higher priced mbs. The Asus Sabertooth is the most expensive but I use that mb for BOTH my 8350 and my 8320 and it is built like a tank.

The 3570k is not as sensitive to the quality of mb for OCs ( it needs a better heatsink!). You could actually buy a less expensive 3570k bundle and likely get better gaming performance.

I notice from your sig you are well aquainted with high end computer parts so you will have the skill and knowledge to make the correct choice for you.

Good luck with the build and keep us posted with your results.

Your' right about the i5 3570K needing a better heatsink. I got a big honking cooler on it and it still would hit over 50C when on load.

However recently I got a great deal on the Xeon chip so sold the i5. Even the new Ivy Variant Xeon would go over 50C but as the Thermal paste cured it not doesn't go over 45C when on load. This is running passively but does have a 140MM fan sucking out hot air from the cpu area.

As for the i5 3570k for the new build, I am going to wait until Haswell comes out and BF sales In the meantime my bro could use a new GPU instead. He has an old but true 4870. Probably get him an Nvidia due to power draw and heat issues with his case.

 

[guskline]

The sad thing about the 8350? It should have been the Bulldozer. AMD dropped the ball releasing the 8150 (I had one and replaced it with a 8350!).

 

[Bill Brasky]

I thought we had moved beyond average FPS as a valid performance measurement.

There should have at least been minimums.

Also, at least according to the Steam survey, less than 1% of people game at 1440.

This. Average fps simply does not correlate with smoothness. I'm not trying to take anything away from AMD, because they make great gaming machines. But in this day and age, minimums and time plots are the only performance data I really pay attention to with regards to gaming.

 

[spaceman]

always been very pleased w. my amd cpus
bunch of synth test apparently verifies this

 

[Idontcare]

I'm impressed with the FX-8350 as a gaming chip overall, but I'm not impressed with it in these benchmarks for the simple reason that if you're going to spend the money on a multi-GPU setup, you'd be dumb not to spend the extra money on the fastest CPU available to fuel them.

I do feel though that the power consumption difference between the 8350 and the 3770K is much less of a factor in a multi-GPU setup considering it's the GPU's that are using the majority of the system power.

It is a classic catch-22 though in terms of logic versus fallacy of logic.

If one argues that the cost associated with the power-consumption delta is irrelevant then one must concede so too is the price differential between the two processors and as such one would opt to purchase the more expensive and higher performing processor.

You either argue that one doesn't care about the smallish price differences and thusly one should logically opt to purchase the 3770k to maximize the ROI of their multi-GPU investment, or you can argue that the price delta is a big deal and as such the cost of electricity in running the 8350 versus the 3770k is of material significance and one ought to buy the 3770k to lower their TCO.

In either case, the conclusion is the same though - the 3770k is what one should buy to pair with their multi-gpu setup if the choice is limited to exclude S2011 options.

 

[Charles Kozierok]

You either argue that one doesn't care about the smallish price differences and thusly one should logically opt to purchase the 3770k to maximize the ROI of their multi-GPU investment, or you can argue that the price delta is a big deal and as such the cost of electricity in running the 8350 versus the 3770k is of material significance and one ought to buy the 3770k to lower their TCO.

Every Intel v. AMD thread does seem to boil down to this, does it not?

 

[ElFenix]

It is a classic catch-22 though in terms of logic versus fallacy of logic.

If one argues that the cost associated with the power-consumption delta is irrelevant then one must concede so too is the price differential between the two processors and as such one would opt to purchase the more expensive and higher performing processor.

You either argue that one doesn't care about the smallish price differences and thusly one should logically opt to purchase the 3770k to maximize the ROI of their multi-GPU investment, or you can argue that the price delta is a big deal and as such the cost of electricity in running the 8350 versus the 3770k is of material significance and one ought to buy the 3770k to lower their TCO.

In either case, the conclusion is the same though - the 3770k is what one should buy to pair with their multi-gpu setup if the choice is limited to exclude S2011 options.

$130 price difference at newegg right now between 3770k and 8350.

lets say you game 6 hours a day. that's a lot for anyone who actually works a job and/or has a family or gf. we'll also pretend that somehow GPU limited gaming also sucks down the same CPU power as the x264 first pass, which gives us a power difference of 75 watts per anand's vishera review. in both cases we're probably overstating the difference in favor of the 3770k

that's 450 watt hours per day. just over 3 KWh each week at 3150, or 163.8 KWh per year. electricity is about 11 cents a KWh right now, here. that's an $18.18 difference a year.

iow, it'd take you 7 years to make up the difference in initial price due to electric costs.

 

[Chiropteran]

iow, it'd take you 7 years to make up the difference in initial price due to electric costs.

Exactly. Often it's even longer, because most games will only fully load 1-2 cores at best.

 

[UNhooked]

In either case, the conclusion is the same though - the 3770k is what one should buy to pair with their multi-gpu setup if the choice is limited to exclude S2011 options.

I would have to respectfully disagree with you. If one is building a budget multi gpu setup the difference in cost between the CPus can make all the difference in what GPUs one might be able to afford.

Of course if price is no object then the above statement becomes moot.

That being said all this time the assumption was Intel is the way to go if you want bleeding edge gaming performance. In my book this review shows that for a budget minded user, all is not lost going AMD route. And that AMD isn't that bad as the general public make it out to be. Losing a few FPS in gaming isn't going to take away from overall gaming experience

 

[BUnit1701]

I am confused as to why GTX 680s or GTX 670s were not used instead of 580s? Good Lord the 780s may be out soon! On the AMD side they used the latest series of high end video cards. Doesn't this lack of testing the Nvidia 600 series cards (should be GTX680) call into question the results? I am not pro Nvidia or Pro AMD. However, when an article is written in April 2013 by a testor who uses 4 7970s vs 2 580s isn't it logical to ask "Hey why no 680s or even 670s?"

This is addressed in the article, he has a limited budget for GPUs (as he is the motherboard reviewer) so he alternates generations, he bought 2 580s, then 4 7970s, he will buy 780s next year.

 

[frozentundra123456]

I would have to respectfully disagree with you. If one is building a budget multi gpu setup the difference in cost between the CPus can make all the difference in what GPUs one might be able to afford.

Of course if price is no object then the above statement becomes moot.

That being said all this time the assumption was Intel is the way to go if you want bleeding edge gaming performance. In my book this review shows that for a budget minded user, all is not lost going AMD route. And that AMD isn't that bad as the general public make it out to be. Losing a few FPS in gaming isn't going to take away from overall gaming experience

I dont think many people still say that the FX is terrible. Is it the best choice, however?

If you want to compete on price, get the 3570k. A very minimal price difference, better gaming in the vast majority of games, and lower power usage. If you want absolute performance, get the 3770k, although it is more expensive. The FX is perfectly adequate, I just dont see the case for it for a gaming oriented rig. Most choices are a trade off. You give up one thing to get another. I just dont see that trade off in the 3570k vs 8350 for gaming. I dont see any real advantage to the 8350, although it is perfectly competent.

 

[jaqie]

Re TCO, power cost, and initial investment.

When I got my thuban 6 core, I literally spent every dime I had on my upgrade. $125 on the cpu was the limit, and I got a hell of a powerful cpu for that price two years ago. The power difference since then has not been great enough that I could have taken out a loan (not that I have good credit anyway) and paid on the interest and saved on electricity enough to pay that.

The original investment is a huge factor to some people, who want the most power they can afford, even if it is at the cost of a dollar or three per month on electric bill.

 

[guskline]

The real problem the 8350 has/had was the literal beating AMD took when it released the overpriced and underperforming 8150. The barrage of bad press so poisoned the field that a lot of reviewers were hesitant to consider the 8350. Conversely, Intel had ( and justifiably so) pretty strong reviews for the 2500k/2600k/2700k. Ivy Bridge improved the performance somewhat and ushered in the 22nm size. With the exception of OCers complaint about higher heat output of the Ivy vs Sandy Bridge Intel has managed to have a very favorable view.

AMD did help itself by lowering the release price of the 8350 and upping the stock clock to 4Ghz. Overclocking is still doable but power draw begins to exponentially climb @4.5-4.6Ghz.

The other real problem for the 8350 will be the release of Haswell. How will AMD respond, if it all?

Personally, the fallacy of the article lies in using 4 7970s. If you can afford 4 of them, surely you can afford a 3770k rig if not a 2011 rig.

 

[ShintaiDK]

$130 price difference at newegg right now between 3770k and 8350.

lets say you game 6 hours a day. that's a lot for anyone who actually works a job and/or has a family or gf. we'll also pretend that somehow GPU limited gaming also sucks down the same CPU power as the x264 first pass, which gives us a power difference of 75 watts per anand's vishera review. in both cases we're probably overstating the difference in favor of the 3770k

that's 450 watt hours per day. just over 3 KWh each week at 3150, or 163.8 KWh per year. electricity is about 11 cents a KWh right now, here. that's an $18.18 difference a year.

iow, it'd take you 7 years to make up the difference in initial price due to electric costs.

Assuming you pay 11cents per Kw/H and not 30 or 40. Not to mention more heat/noise. And that the FX will be obsolete way before due to poor performance. Not to mention the 2500K whoops it as well massively. And thats in close to the same price range. (Today it would be a 3570K).

There is just no excuse to get a FX, besides the irrational.

 

[guskline]

This is addressed in the article, he has a limited budget for GPUs (as he is the motherboard reviewer) so he alternates generations, he bought 2 580s, then 4 7970s, he will buy 780s next year.

Sorry, I'm not "buying" that excuse. Buy 2 7970s and 2 670s or better yet "borrow" some 670s or 680s from another division.

 

[Unoid]

I have two issues with this review.
1) Is a big one - average FPS with a single number. No attempt to look at even minimums or the range of frame rates. Tomshardware did a significantly better review some months ago testing the 8350 and showed while averages were similar there were problems with consistent frame delivery, the top and bottom frame rate delivered was different. Doing a single FPS test in today's tooling is unacceptable because we know the data is often misleading.

2) The range of games used was hardly a broad range of different types of games. In essence what we had was 3 known GPU limited games with very light CPU usage and 1 heavy CPU user. It isn't a sufficient list of games to say much other than that this very odd selection works the way it does.

I have plenty of games that are clearly CPU limited at 4.5Ghz and 2 680's. It is a real struggle to get to 120 fps, and that is often dominated by the CPU not the GPUs.

I agree with you.

I usually run my 2600k at 4.6ghz to keep the volts under 1.3v But I may need to start running 4.8-4.9 in the near future for my 680 SLI.

 

[KingFatty]

Couldn't the article just have concluded that gamers are moving onto 1440p and higher resolutions, and at those resolutions you are GPU-limited anyway (even if running dual GPUs) so it doesn't really matter what your CPU is for most games.

 

[UNhooked]

I dont think many people still say that the FX is terrible. Is it the best choice, however?

If you want to compete on price, get the 3570k. A very minimal price difference, better gaming in the vast majority of games, and lower power usage. If you want absolute performance, get the 3770k, although it is more expensive. The FX is perfectly adequate, I just dont see the case for it for a gaming oriented rig. Most choices are a trade off. You give up one thing to get another. I just dont see that trade off in the 3570k vs 8350 for gaming. I dont see any real advantage to the 8350, although it is perfectly competent.

I agree with you. Going by the prices at my local MC and promos they are very close, that is the i5 3570k and FX-8350. Still if you have an existing AMD rig then the FX- 8350 would drop in nicely.

 

[frozentundra123456]

Sorry, I'm not "buying" that excuse. Buy 2 7970s and 2 670s!

Kind of off topic, but doesn't it seem that there has been a dearth of reviews on AT lately, as far a cpu and gpu performance in some of the new games coming out?

 

[guskline]

Kind of off topic, but doesn't it seem that there has been a dearth of reviews on AT lately, as far a cpu and gpu performance in some of the new games coming out?

Agreed. Probably the expense of the cpus and gpus plus the games.

 

[Idontcare]

Couldn't the article just have concluded that gamers are moving onto 1440p and higher resolutions, and at those resolutions you are GPU-limited anyway (even if running dual GPUs) so it doesn't really matter what your CPU is for most games.

You need to create a little controversy to sell your wares, advertising 101. Nobody is interested in reading a snoozer.

Agreed. Probably the expense of the cpus and gpus plus the games.

Anandtech doesn't buy hardware or software. Can't think of a review site that does anymore. It is an interconnected world of advertisers and sponsorships out there nowadays.

Its nothing like the good old days where reviewers bought retail products the same as you and I only to post a review of them online sans any question of conflict of interest between reviewer and paycheck. Things just don't work like that when you want to scale up the business to a full-time proposition.