How Did Anand Let This Get Published?

[Sheep221]

Well, just to restate my earlier query which has thus far gone unanswered... what exactly does the state of desktop Haswell have to do with a review of mobile Richland?

It's one thing to draw conclusions and provide editorial insight based off the content of a review, it's another to go off on a tangent into a different product segment.

Well that line was not really needed, but he mentioned that in the sense that AMD needs to seriously wake up because it is causing Intel to release [not-so-good] products in general. That was his point. Which is true and probably everyone beside op got this right away.


No profanity in the tech forums, please.

Moderator jvroig

 

[Khato]

Well that line was not really needed, but he mentioned that in the sense that AMD needs to seriously wake up because it is causing Intel to release [not-so-good] products in general. That was his point. Which is true and probably everyone beside op got this right away.

Heh, now Haswell is a '[not-so-good] product'? Just because it doesn't offer anything more than minimal improvements on the desktop over what was already an excellent product? If he wanted to write an editorial on Haswell desktop then why didn't he? Or suggest to Anand that a comment of that nature be put into the desktop Richland vs. Haswell review of a few weeks ago?

Instead he attacks desktop Haswell in a review of mobile Richland in order to distract from the fact that he's ignoring mobile Haswell in his analysis. Sorry, but that's not the quality of journalism I've come to expect from Anandtech.


For future reference, please don't quote profanity in the tech forums. Thank you.

Moderator jvroig

 

[ShintaiDK]

I wonder if this review is another AMD mandated one like Trinity. And the reviewer just bended over.

It could have been written by the kinds of JF.

 

[Sheep221]

Heh, now Haswell is a '[not-so-good] product'? Just because it doesn't offer anything more than minimal improvements on the desktop over what was already an excellent product? If he wanted to write an editorial on Haswell desktop then why didn't he? Or suggest to Anand that a comment of that nature be put into the desktop Richland vs. Haswell review of a few weeks ago?

Instead he attacks desktop Haswell in a review of mobile Richland in order to distract from the fact that he's ignoring mobile Haswell in his analysis. Sorry, but that's not the quality of journalism I've come to expect from Anandtech.

Yes it was inappropriate abit, but it is not something that require complaint. Review is type of publication where author is allowed to express own personal opinions about the reviewed thing. While hardware reviews are generally very constructive, you are allowed to say that this product is crap, or another is better than that crap. It's simple to understand and most of all he was right about haswell.
Doesn't matter that Intel is faster than AMD, yes IVB and Haswell are nothing more than rebadged sandy bridges with lower build quality, which in prevention of overheating, caused many people to do unnecessary steps of delidding, replacing thermal paste and stick the IHS or go bare-die.
I find this to be very crap marketing and while the CPUs are still faster than AMDs, they are crap because they offer no performance improvement over sandy bridge but are marketed like that on the desktop platform.
And most of all, he included haswell based laptop in his review in all benchmarks where he also constructively stated that AMD has to work more on decreasing TDP and increasing performance. There is nothing wrong with the review and it was well written. Some of you people are just generalizing everything and can't deal with Intel being seen in bad light. Because you haven't heard of different company than Intel or Nvidia, respectively.


For future reference, please don't quote profanity in the tech forums. Thank you.

Moderator jvroig

 

[galego]

Yea, this is kind of along the lines of what I was thinking. How can you claim one company is not improving fast enough because they are so far ahead that there is no competition, and then criticize that company for its performance.

Where does that leave AMD? If haswell is a joke, what is Vishera, a worse joke?

If Haswell is Intel Bulldozer then Vishera is Haswell 2.0

 

[guskline]

Vishera is what Bulldozer should have been.

We can scream all we want but Intel has focused the mid level desktop chips (Haswell) on more versatility with power savings. I've owned a Sandy Bridge 2500k and now own a Ivy Bridge 3770k. I also owned an AMD 1100T, Bulldozer 8150 and now own a PileDriver 8350 and 8320. The power savings and memory utilization of Intel is much better than AMD AND the CPU horsepower is better. This is based on my testing of the chips I've owned.

Intel has the luxury of high end enthusiast chips. Sandy Bridge E socket 2011. What does AMD have? PileDriver. On almost every test my 3770k outperforms my 8350. I can only imagine the disparity when the Sand Bridge E comes into play.

I think Haswell ( I don't own one yet) is more of a modular design for the future that will enable Intel to effectively compete in the mobile and net book market.

 

[Pandamonia]

Gotta love the thought processes around here.

AMD is still battling Core2 for IPC, meanwhile Intel lands another 20% IPC increase tock, lower power consumption, great scaling up and down with clocks and voltage, as well as adding new instructions.

:thumbsup:

Show me this 20% IPC gain from Ivy please.

 

[CHADBOGA]

Show me this 20% IPC gain from Ivy please.

He referenced from Tock to Tock, so he means from Sandy.

 

[Pandamonia]

He referenced from Tock to Tock, so he means from Sandy.

I still dont think the figure is as high as 20%.

Average from AT is around 12-15% They are dreaming if they think haswell is 20% faster than sandy.

Then deduct the lower overclocks and basically we are likely to have gone backwards or stood still.

 

[Sable]

The real question is How Did Intel Let Haswell Get Released?

Even assuming that Haswell was a major step in the mobile direction... it was two steps backward in the desktop: (1 - 2) = -1

HEY YOU SHOULD SIG THAT!! IT'S TOTALLY LOLZ!!!

How is desktop Haswell 2 steps backwards?! Did they lower IPC? Did they increase power consumption? Please explain.

 

[Pandamonia]

HEY YOU SHOULD SIG THAT!! IT'S TOTALLY LOLZ!!!

How is desktop Haswell 2 steps backwards?! Did they lower IPC? Did they increase power consumption? Please explain.

Lowered overclocking. Stopped overclocking on non K SKU's

Added more thermal problems for desktop overclockers.

Haswell is a step back for enthusiasts. Not for many others.

 

[JimmiG]

The net result is basically plus/minus 0. It overclocks 10-15% less and is 10-15% faster per clock.

 

[Arachnotronic]

Lowered overclocking. Stopped overclocking on non K SKU's

Added more thermal problems for desktop overclockers.

Haswell is a step back for enthusiasts. Not for many others.

"Lowered overclocking"? You say this as though the company flipped a switch to screw up overclocking. No, the silicon was targeted for low power consumption (yes, Haswell cores fit in tablets), and the fact that it can scale to ~4.5GHz is mind-blowing. Further, with the IPC increase, the majority of the world - who run chips at stock - saw a performance increase. It gets better when software is recompiled.

It might be a "step back" for Enthusiasts, but remember that this isn't the enthusiast part; enthusiast refresh is right around the corner with IVB-E and then HSW-E is coming a year after that. I guarantee a 6 core IVB beats the snot out of a Haswell 4C, and I'm even more confident an 8C haswell will beat up the "Haswell Refresh" part.

 

[Pandamonia]

"Lowered overclocking"? You say this as though the company flipped a switch to screw up overclocking. No, the silicon was targeted for low power consumption (yes, Haswell cores fit in tablets), and the fact that it can scale to ~4.5GHz is mind-blowing. Further, with the IPC increase, the majority of the world - who run chips at stock - saw a performance increase. It gets better when software is recompiled.

It might be a "step back" for Enthusiasts, but remember that this isn't the enthusiast part; enthusiast refresh is right around the corner with IVB-E and then HSW-E is coming a year after that. I guarantee a 6 core IVB beats the snot out of a Haswell 4C, and I'm even more confident an 8C haswell will beat up the "Haswell Refresh" part.

Sorry i didnt know a 2 year old architecture based on 32nm was considered the "enthusiast" chip in the intel product portfolio.

Also intel lowered overclocking by using TIM and poor spacing between the IHS and the cores. This is lazy mass production to reduce costs to maximise profits.

You should take off those Intel Goggles you have on because they are blinding you from what everyone else realises.

 

[StinkyPinky]

This is what happens when there's no competition. I'm not sure what people were expecting. Of course Intel are coasting....how naive are people to think they would push themselves when there's no reason too.

 

[krumme]

They are not coasting but harvesting. The market is regarded as a cash cow soon to be a dying dog.

Well, thats how Intel is treating it. So when the desktop market is dead, they can sell Haswell #3 for 200usd for tablets and other ultra thin clients. Good luck.

 

[ShintaiDK]

I guess the record R&D budgets and factory build is just coasting along.

 

[frozentundra123456]

Yes it was inappropriate abit, but it is not something that require complaint. Review is type of publication where author is allowed to express own personal opinions about the reviewed thing. While hardware reviews are generally very constructive, you are allowed to say that this product is crap, or another is better than that crap. It's simple to understand and most of all he was right about haswell.
Doesn't matter that Intel is faster than AMD, yes IVB and Haswell are nothing more than rebadged sandy bridges with lower build quality, which in prevention of overheating, caused many people to do unnecessary steps of delidding, replacing thermal paste and stick the IHS or go bare-die.
I find this to be very crap marketing and while the CPUs are still faster than AMDs, they are crap because they offer no performance improvement over sandy bridge but are marketed like that on the desktop platform.
And most of all, he included haswell based laptop in his review in all benchmarks where he also constructively stated that AMD has to work more on decreasing TDP and increasing performance. There is nothing wrong with the review and it was well written. Some of you people are just generalizing everything and can't deal with Intel being seen in bad light. Because you haven't heard of different company than Intel or Nvidia, respectively.


For future reference, please don't quote profanity in the tech forums. Thank you.

Moderator jvroig

Actually the "review" sounded almost more like a forum post rather than a true review. Even ignoring the off topic rant against desktop haswell, the real problem was that he compared the highest end full power chip from amd against a low power ultrabook chip from Intel. It was by far the worst review I have ever seen on AT.

 

[FwFred]

Actually the "review" sounded almost more like a forum post rather than a true review. Even ignoring the off topic rant against desktop haswell, the real problem was that he compared the highest end full power chip from amd against a low power ultrabook chip from Intel. It was by far the worst review I have ever seen on AT.

I wonder system availability is a problem for him? I'm not sure he has access to many systems to compare it to. Still especially for GPUs which scale very well with power budget (if wide enough), it's just incomplete to be comparing 15W vs. 35W.

 

[JimmiG]

It might be a "step back" for Enthusiasts, but remember that this isn't the enthusiast part.

4770K pricing is borderline "enthusiast", though. Clearly, the lack of competition allowed Intel to jack up prices. If AMD FX was truly competitive, I think the 4770K would have been a $200 to $239 part and the 4670K a $150 one. It's also about time we got a 6-core mainstream desktop part. Quad-Core was the high-end in 2008, but this is 2013. Even my phone has a quad-core CPU.

Compare this to 2000 - 2005 - we went from the Athlon 1 GHz to the Athlon 64 X2 2.5 GHz. Twice the number of cores, 2.5x the clock speed, significant IPC improvements and a new 64-bit architecture.

Second to pricing, the IHS issue is really the biggest concern. You shouldn't have to pay $500+ for IB-E to get something that was standard on all CPUs a few years ago. There are really no excuses that can justify that. This "heat chamber" under the IHS *is* the main reason HW does not overclock as high and runs so hot. You just can't dial in the voltages most CPUs need for 4.6+ GHz without reaching 100+C.

 

[mrmt]

Maybe that reviewer should ask AMD where's Vishera successor, or where's the 20nm process that AMD will use in their high clock parts, or why AMD is bringing Steamroller as APU first (and maybe APU only, which suits perfectly the mobile market).

 

[frozentundra123456]

I wonder system availability is a problem for him? I'm not sure he has access to many systems to compare it to. Still especially for GPUs which scale very well with power budget (if wide enough), it's just incomplete to be comparing 15W vs. 35W.

I am sure the problem was availability. That still does not make the results any less invalid. He should have just compared Richland to a full power mobile ivy i5 and i7. Top of the line mobile chip for both.

 

[frozentundra123456]

4770K pricing is borderline "enthusiast", though. Clearly, the lack of competition allowed Intel to jack up prices. If AMD FX was truly competitive, I think the 4770K would have been a $200 to $239 part and the 4670K a $150 one. It's also about time we got a 6-core mainstream desktop part. Quad-Core was the high-end in 2008, but this is 2013. Even my phone has a quad-core CPU.

Compare this to 2000 - 2005 - we went from the Athlon 1 GHz to the Athlon 64 X2 2.5 GHz. Twice the number of cores, 2.5x the clock speed, significant IPC improvements and a new 64-bit architecture.

Second to pricing, the IHS issue is really the biggest concern. You shouldn't have to pay $500+ for IB-E to get something that was standard on all CPUs a few years ago. There are really no excuses that can justify that. This "heat chamber" under the IHS *is* the main reason HW does not overclock as high and runs so hot. You just can't dial in the voltages most CPUs need for 4.6+ GHz without reaching 100+C.

I could agree to a certain extent with most of what you are saying, but every tech product makes much faster performance gains in the early part of its life than after the tech has become mature, so it is not realistic to expect the type of gains we saw back then to continue.

 

[blackened23]

4770K pricing is borderline "enthusiast", though.

Uh, no, it's not. The 300$ price range has always been for mainstream CPUs dating back even two decades ago. Performance/enthusiast CPUs have *always* been substantially more expensive, period. The pentium-150 cost nearly 600$ at release in 1996 IIRC. And here people are complaining about a cheap CPU such as the 4770k? 300 bucks? Get real. I think even the Celeron 300a way back in the day cost nearly 300$. And that was the lowest of the low end back then.

The 4770k is a mainstream part. Not enthusiast. See the 3930k and soon to be 4930k for the real enthusiast part, and priced accordingly.

 

[galego]

HEY YOU SHOULD SIG THAT!! IT'S TOTALLY LOLZ!!!

How is desktop Haswell 2 steps backwards?! Did they lower IPC? Did they increase power consumption? Please explain.

Consider 3770k vs 4770k:

• 1-5% increase in average performance, but performance regressions in specific tests.
• Higher power consumption.
• Run hotter.
• Poor OC.

And that is the short list.