Discussion i7-11700K preliminary results

[GunsMadeAmericaFree]

early benchmarks

more benchmarks

Just thought I would provide these so folks could draw their own conclusions. Evidently the processors have been getting sold for a few days now in Germany.

 

[Zucker2k]

Did you read the article NDA was not broken.

The review sample they have from intel will be published on the 30th.

They don't need to publish that anymore. Is the retail chip different from the chip Intel sent them? The retail chip is what everyone's going to get. If they want to review the 11700k on a different motherboard then they should title it as such. As far as I'm concerned, they've already taken their due bite out of the RKL-S review cake.

 

[linkgoron]

Once the magical Vega Rocket Lake drivers appear, we'll see much better numbers!

This is exactly it. Why should Intel lend any kind of legitimacy to this ill-conceived move when they had already made their position clear that they disapprove of what Mind Factory did? This was just a bad call but it's good in a way, because now everyone knows Anandtech will not hesitate to ignore NDA given the opportunity.
There are lots of German reviewers who probably bought the chips but are holding off because they signed NDAs and knew to take the high road. We all know if this was AMD hardware being "reviewed" on non final motherboard and bios with sketchy details, all hell would've broken loose by now. The double-standards is just sickening.

Nobody would care if the numbers were good, but the Intel numbers are disappointing - so people are angry. Anandtech did nothing wrong. They bought the CPU at a store, and benchmarked it.

 

[Zucker2k]

NDA cannot be broken on a part purchased in the retail channel. NDA never applied to the chip Ian bought.

The chip was sold without the approval of Intel. The whole world knew about this so why would a professional reviewer exploit this loophole because of the technicality that arose out of an illegality, and an NDA signee at that?

 

[Makaveli]

They don't need to publish that anymore. Is the retail chip different from the chip Intel sent them? The retail chip is what everyone's going to get. If they want to review the 11700k on a different motherboard then they should title it as such. As far as I'm concerned, they've already taken their due bite out of the RKL-S review cake.

I guess we are all thankful that isn't a decision you get to make for the site. I'm waiting for the full review on the 30th which should go into more detail about memory performance for intel's gear down mode.

 

[Panino Manino]

Did you read the article NDA was not broken.

The review sample they have from intel will be published on the 30th.

NDA cannot be broken on a part purchased in the retail channel. NDA never applied to the chip Ian bought.

This is exactly it. Why should Intel lend any kind of legitimacy to this ill-conceived move when they had already made their position clear that they disapprove of what Mind Factory did? This was just a bad call but it's good in a way, because now everyone knows Anandtech will not hesitate to ignore NDA given the opportunity.
There are lots of German reviewers who probably bought the chips but are holding off because they signed NDAs and knew to take the high road. We all know if this was AMD hardware being "reviewed" on non final motherboard and bios with sketchy details, all hell would've broken loose by now. The double-standards is just sickening.

Thinking more about this, besides the CPU he bought, Ian have another one in his possession gave to him by Intel? If he has he can test the two and compare.
Other thing, the people who writes for Anantech have contacts, right? They may know more about a product than what Intel discusses behind the NDA. Her may be wrong and performance may change a bit like it happened in recent past with Ryzen, but he must have a very good bet that this is it and his review will be proved right.

Rather than just "publishing a review" and Ian/Anandtech did here is more akin to journalism work. Tech websites mustn't exist just to be mouth pieces for these companies, they also should question and investigate. See, how many people are 100% sure that he is wrong "because Intel promised better performance"? They're question this review because "this mustn't be right", why they think this way?

 

[Makaveli]

See, how many people are 100% sure that he is wrong "because Intel promised better performance"? They're question this review because "this mustn't be right", why they think this way?

Those that are bias will see it where none exist. For the rest of us we can see the forest for the trees.

 

[UsandThem]

I don't think Ian violated the Intel NDA based on law or anything where Anandtech could be sued for breach of contract, etc. I do think it was in bad form from a generally well-respected hardware review site.

However, I do think that Intel will remember this for future CPU launches, and Anandtech will likely get their sample much later than other review sites so their review is delayed by a significant amount of time.

Now I'm sure Anandtech could buy a retail CPU like they did here (maybe from a company who sells them early), or maybe they will only find one after they are officially launched in the retail channel. I don't think much will change concerning the results (it looks like Intel's new CPU is a major disappointment), but I think Anandtech should have not published their review earlier than when the NDA expired

 

[ondma]

Did you read the article NDA was not broken.

The review sample they have from intel will be published on the 30th.

Semantics. Perhaps, technically, the letter of the NDA was not violated, but the spirit/intent of the NDA most certainly was.

 

[Makaveli]

I don't think intel will change anything about sending samples to this site its too big. Secondly, it seems to me there are a lot of users here and on the internet that just wants reviews to show the results they want to see, and that is not how this works.

 

[gdansk]

In the end retail units were out early. This thread was about results from a number of sites and forums. I would rather have Ian and Andrei run some tests than get unreliable numbers from random forum posters (no offense).

The unit provided by Intel will be reviewed as well, I'm sure. And we can compare its performance at that point. If HWUB has had Rocket Lake chips for sometime, I assume Ian does as well. And I doubt he'd publish this review if the results differed greatly. If there is a wonderful firmware update, then a bunch of people will be redoing their testing in the next few weeks.

 

[CHADBOGA]

Leaving aside the issues of whether a new bios will improve performance or not, why does performance on Cinebench improve so much more than most other applications?

Does this now make Cinebench a poor benchmark?

 

[Thunder 57]

This is exactly it. Why should Intel lend any kind of legitimacy to this ill-conceived move when they had already made their position clear that they disapprove of what Mind Factory did? This was just a bad call but it's good in a way, because now everyone knows Anandtech will not hesitate to ignore NDA given the opportunity.
There are lots of German reviewers who probably bought the chips but are holding off because they signed NDAs and knew to take the high road. We all know if this was AMD hardware being "reviewed" on non final motherboard and bios with sketchy details, all hell would've broken loose by now. The double-standards is just sickening.

I'm pretty sure hell has broken loose, plenty of people have been complaining about it. If RL had put up good numbers then no one would give a damn. There's your double standard. People are just surprised it is such a turd sandwich and it makes no sense with ADL coming so soon.

Leaving aside the issues of whether a new bios will improve performance or not, why does performance on Cinebench improve so much more than most other applications?

Does this now make Cinebench a poor benchmark?

Pretty sure the rule is if Intel does well it is a "real world" benchmark. They certainly didn't have a problem with Cinebench when they were stomping AMD. In all seriousness though, maybe it likes the new cache structure, particularly the L2?

 

[Hulk]

Leaving aside the issues of whether a new bios will improve performance or not, why does performance on Cinebench improve so much more than most other applications?

Does this now make Cinebench a poor benchmark?

Yes, Rocket Lake does show a nice bump across every version of Cinebench tested compared to Comet Lake. Unfortunately the 5800X still beats it in every single Cinebench version, both ST and MT. So to answer your question, no, Cinebench is still the same benchmark it was last week.

 

[ondma]

I'm pretty sure hell has broken loose, plenty of people have been complaining about it. If RL had put up good numbers then no one would give a damn. There's your double standard. People are just surprised it is such a turd sandwich and it makes no sense with ADL coming so soon.



Pretty sure the rule is if Intel does well it is a "real world" benchmark. They certainly didn't have a problem with Cinebench when they were stomping AMD. In all seriousness though, maybe it likes the new cache structure, particularly the L2?

Pretty sure the "double standard" works both ways, i.e. AMD fans assume Intel cheats no matter what.

 

[Panino Manino]

Semantics. Perhaps, technically, the letter of the NDA was not violated, but the spirit/intent of the NDA most certainly was.

What spirit of NDA?
The NDA is more about holding back the information that Intel gives ahead of the launch and control the date of the spread of this information. Ian din't gave us anything that Intel gave to him.

Again, I get the feeling that this is only being discussed among us, the public, because the results are bad and some of us were hoping otherwise. Even on 14nm this is still a new architecture that should had crushed Zen 3, I myself was expecting a very different result. Some are not taking this break in expectations very well and are venting their frustration on Ian.
Let's hope that when the launch day comes these people will have calmed down because again I say that Ian must have a very good bet on his hands. With the information that Intel already gave to him he should have a better understanding of what is causing the regressions and will have plenty of time to challenge Intel's marketing claims when the official review day comes. This review is just part 1. Can you imagine other site having to trouble (and skills) to go in deep on the details of this CPU and explain in a way that common user can understand like Anandtech does?
I should be glad that I'll not be overwhelmed having to read 3 thousand pages of information and benchmarks at once.

 

[Panino Manino]

Pretty sure the "double standard" works both ways, i.e. AMD fans assume Intel cheats no matter what.

But it's true and is common knowledge.
Don't misunderstand, it's not Intel "cheating" everywhere, is just that they have the opportunity to make things difficult for AMD and they do. You don't need to believe in me or in "AMD fans", the Courts believe.
This discussion was closed years ago, but to be fair that problem improved a lot and it's not that relevant anymore.

 

[majord]

No NDA broken, but for respect of relationship with Intel, perhaps should have been called a Preview, and perhaps been less detailed.

Keeps referring to Anand's early Sandy bridge "review" to justify why it's all OK, but that wasn't a review, it was clearly titled a 'Preview' , and clearly stated the results were Preliminary, and could or would likely change. Quite different.

 

[tamz_msc]

No NDA broken, but for respect of relationship with Intel, perhaps should have been called a Preview, and perhaps been less detailed.

Keeps referring to Anand's early Sandy bridge "review" to justify why it's all OK, but that wasn't a review, it was clearly titled a 'Preview' , and clearly stated the results were Preliminary, and could or would likely change. Quite different.

For me the crucial part is that Intel didn't care to comment on whether Anandtech could publish their review. If there was indeed a firmware update in the works that could actually alter the results in a significant way, don't you think that Intel would have reached out to Ian and say "don't publish it now, we have a BIOS update in the works and that the performance on current BIOS isn't final/representative of what you'll see on launch day"?

Nothing of that sort happened, which is why the comparison with Sandy Bridge situation is not applicable, in my opinion.

 

[majord]

For me the crucial part is that Intel didn't care to comment on whether Anandtech could publish their review. If there was indeed a firmware update in the works that could actually alter the results in a significant way, don't you think that Intel would have reached out to Ian and say "don't publish it now, we have a BIOS update in the works and that the performance on current BIOS isn't final/representative of what you'll see on launch day"?

Nothing of that sort happened, which is why the comparison with Sandy Bridge situation is not applicable, in my opinion.

I certainly do agree with that. and in the absense of that , I don't think they had anything else to say really.. They're not going to tell him 'that's poor form' , but they may still take it that way.

 

[CHADBOGA]

Yes, Rocket Lake does show a nice bump across every version of Cinebench tested compared to Comet Lake. Unfortunately the 5800X still beats it in every single Cinebench version, both ST and MT. So to answer your question, no, Cinebench is still the same benchmark it was last week.

My point is why does Cinebench R20 single core score improve by 11% over the 10700K, but either regress or have a much smaller performance increase in so many other applications when you compare the 11700K to the 10700K?

 

[UsandThem]

I certainly do agree with that. and in the absense of that , I don't think they had anything else to say really.. They're not going to tell him 'that's poor form' , but they may still take it that way.

I think that sort of thing is lost on a lot people.

Just because you "can" do something, it doesn't always mean you "should" do it.

I don't take any kind of pity on Intel.......they've have had their fair share of brain farts over a long period of time, I just think the decision to publish the review early by obtaining a retail CPU sold before it should have been sold, is something that someone in the Future corporation hierarchy to pause and think "maybe we shouldn't". I personally believe this is going to negatively affect Anandtech's relationship with Intel, and may result in less one-on-one interviews and possibly could spill over into the review kits in the future.

 

[Kocicak]

The chip was sold without the approval of Intel. The whole world knew about this so why would a professional reviewer exploit this loophole because of the technicality that arose out of an illegality, and an NDA signee at that?

I will fix it for you: The chips was were sold without the approval of Intel.

Have you noticed, that more people who got their CPUs early have been publishing benchmarks? In this situation, why not to make a proper review?

Speaking about exploiting loopholes - have you noticed that there is a whole consulting industry around exploiting legislative loopholes to "optimise taxes"? Companies do the maximum legally allowed to maximise their well being.

Why a reviewer or media company should not do the same?

 

[ondma]

What spirit of NDA?
The NDA is more about holding back the information that Intel gives ahead of the launch and control the date of the spread of this information. Ian din't gave us anything that Intel gave to him.

Again, I get the feeling that this is only being discussed among us, the public, because the results are bad and some of us were hoping otherwise. Even on 14nm this is still a new architecture that should had crushed Zen 3, I myself was expecting a very different result. Some are not taking this break in expectations very well and are venting their frustration on Ian.
Let's hope that when the launch day comes these people will have calmed down because again I say that Ian must have a very good bet on his hands. With the information that Intel already gave to him he should have a better understanding of what is causing the regressions and will have plenty of time to challenge Intel's marketing claims when the official review day comes. This review is just part 1. Can you imagine other site having to trouble (and skills) to go in deep on the details of this CPU and explain in a way that common user can understand like Anandtech does?
I should be glad that I'll not be overwhelmed having to read 3 thousand pages of information and benchmarks at once.

I am sure other sites will do exactly that ---- when the official release date arrives.

 

[Timorous]

No NDA broken, but for respect of relationship with Intel, perhaps should have been called a Preview, and perhaps been less detailed.

Keeps referring to Anand's early Sandy bridge "review" to justify why it's all OK, but that wasn't a review, it was clearly titled a 'Preview' , and clearly stated the results were Preliminary, and could or would likely change. Quite different.

Did you cross check the sandy preview and review results? In the cases where the same tests were done performance was very similar between the 2400 preview part and the 2500K. That was with 5 month to work on updating the BIOS to try and eek out more performance. So while it was stated that they could change there was not much movement.

The chip can be purchased at retail. If an outlet gets one of those retail chips it is fair game. All this they should be fair to other reviewers or they should be fair to Intel is BS. What about consumers in this. The part can (could? not checked today) be purchased at retail. Rather than allowing people to buy them blind Ian purchased one and benchmarked it so you can go in eyes open. That is the service we expect from the tech press and we don't like it when companies have reviews and sale go live at the same time so why all the complaining in this case.

The only reason I can think of as to why people are moaning is because the product is underwhelming. If it was a knockout product and it was faster than the 5800X across the board I bet the response would be very different. (EDIT: Actually this is wishful thinking, instead the AMD fanboys would be complaining for some reason. Probably talking about BIOS updates to fix a security flaw that is going to tank performance so same same but different).

 

[Zucker2k]

The part can (could? not checked today) be purchased at retail. Rather than allowing people to buy them blind Ian purchased one and benchmarked it so you can go in eyes open.

Mind Factory stopped sales many days ago so what exactly is the purpose of posting reviews of a product that's under general NDA? I believe there's a two week gap, iirc, between reviews and when the chips go on sale. That's a long period for a potential customer to gather all the information they need before plunging in.

My point is why does Cinebench R20 single core score improve by 11% over the 10700K, but either regress or have a much smaller performance increase in so many other applications when you compare the 11700K to the 10700K?

Memory and cache latencies don't seem to significantly impact a throughput-oriented bench like CB, all things considered.