Looks pretty much just as I suspected..Intel's chip lineups are pretty consistent. The addition of the 6 core / 12 thread chip from Cannon Lake seems early for an X series but we don't know much about its architecture (just like early days of Ryzen) The thing that I'm most excited about is that Threadripper will have 60 PCIe lanes directly connected to the CPU (and four more going to the chipset). Intel's going to have to bump up their "low end" i7-7800X and i7-7820X if they want to sell a single one. The bottom two chips have been gimped with 28 lanes for at least the two previous chips. Between NVMe drives, GPU, maybe a 10g NIC...I can't be running stuff over the chipset on it's shared x4 lane. A single NVMe drive will saturate a PCIe x4 lane, so they need their own link to the CPU. Having multiple drives plus a GPU makes extra lanes a necessity. Hopefully this kinda trickle down works 🙂

Linus Tech Tips I have some things to say - Core i9 & X299 https://www.youtube.com/watch?v=TWFzWRoVNnE Edit- the "new" X299 is pure crap! If I want more cores I will buy the dual 2011-v3 and would get those: https://www.newegg.com/Product/Product.aspx?Item=N82E16819117632 https://www.newegg.com/Product/Product.aspx?Item=N82E16813182967 And I would get much better(more Stable!) workstation with 24C!. P.S: why we need anothere new Socket for "Up-To" 18 Cores while we have the 2011-V3 socket X99 boards that support Up-To 22 cores Intel Xeon E5-2699-v4 Just another $$$$ move by Intel:bang:.

Mr. Hilbert, when you are reviewing these jokes of HEDT wannabe CPU, please, delid one of them to show the world how Intel ****s on everyone's face with the thermal paste between the IHS and the core... on a 1000+ euros HEDT platform I am a Intel fanboy by the way, and these days i hated myself for being one

This thread is a reminder of why I rarely come to G3D anymore. Carry on.

to be fair this is how it always was when AMD was actually relevant. its been one sided for almost a decade which is why we all forgot

Everyone keeps saying "wait for benchmarks before deciding" but to me, I find both products to be ridiculous, and not in a good way. They get people to approach problems in a more lazy way - rather than try to optimize software to work the best way it can, people instead are like "meh, just throw more cores at it". All these products do is cater to people's impatience, egos, and wastefulness. If these were server chips, I wouldn't care - servers have a need for this many threads.

Its the CPU makers fault also. They went for MHz, that didnt go as far as they hoped. Then they went more cores, and thats sort of coming to a head now. I think the next step will be beefier cores (and HBM on chip), but not going above 8/16 for consumer level chips.

Mr. Hilbert, when you are reviewing these jokes of HEDT wannabe CPU, please, delid one of them to show the world how Intel ****s on everyone's face with the thermal paste between the IHS and the core... on a 1000+ euros HEDT platform I am a Intel fanboy by the way, and these days i hated myself for being one

This is getting old. The reason was already listed because of long term risk using solder, there have been cases of failure. Remember, unlike your wish to have an enthusiast system sitting nicely in your air-conditioned house, Intel has to be compatible to a wide range of temperatures, shipping tolerances and humidity levels, including for long period storage. Most people do not have the luxury of reapplying TIM as needed UNDERNEATH an IHS as you so wish. Just have a look at some of the temperature, electrical and shipping tolerances, etc., that the CPU's have to meet: http://www.intel.com/content/www/us/en/processors/core/7th-gen-core-family-desktop-s-processor-lines-datasheet-vol-1.html Aftermarket thermal paste has the luxury of not having to meet a large fraction of those specifications. In any case, testing has shown that Intel's thermal paste is right up there with the best aftermarket solutions: [spoiler]http://images.anandtech.com/doci/8227/1b%20CPUIHS.png[/spoiler] And just as a point of comparison, the Noctua NT-H1 thermal paste quoted above has the following specifications:

Recommended storage time (before use) up to 2 years Recommended usage time (on the CPU) up to 3 years http://noctua.at/en/nt-h1/specification

Now if you think for a second that Intel could ship a CPU with a three year operating life and that the whole internet wouldn't explode... There are probably a few thousand of people on the internet in total that even care about this out of the billions of processors shipped anyways. Sorry, this is a non-issue.

Mr. Hilbert, when you are reviewing these jokes of HEDT wannabe CPU, please, delid one of them to show the world how Intel ****s on everyone's face with the thermal paste between the IHS and the core... on a 1000+ euros HEDT platform I am a Intel fanboy by the way, and these days i hated myself for being one

It was already done: youtube.com/watch?v=I1Bv8Mxnnlc There is also talk about TIM under the hood. My personal take on this is a little bit different. I do agree that cost per chip switching from TIM to soldering is small, but what needs to be taken into account if fabrication process. Applying TIM instead of soldering HIS is much simpler and, this is IMHO key point, is much safer. I do suspect that failure rate when using TIM is much smaller then soldering so yield per 1k unit is probably much greater making it more $ efficient. Side effect here is increase in thermals, but within safe range, although more in upper spectrum, making it harder to OC using WC or AC solutions, so customers will most likely purchase next iteration sooner then it was before (due to smaller top clock speeds). Additionally it will force to de-lidding and personally I don't see many who would attempt it on 1k+ CPU, not to mention that switching to i.e. liquid metal require maintenance at least once per 1-2years and I personally wouldn't want to bother with whole process again. Of course this is My personal take on this, I'm not metallurgy expert and may be completely wrong why Intel does it. Also it seams like 14 ->18c SKU's will hit market in 2018 so AMD will have nice head start with their TR launch. PS. I was late to reply it seams.

Time for AMD to shine.

It was already done: youtube.com/watch?v=I1Bv8Mxnnlc There is also talk about TIM under the hood. My personal take on this is a little bit different. I do agree that cost per chip switching from TIM to soldering is small, but what needs to be taken into account if fabrication process. Applying TIM instead of soldering HIS is much simpler and, this is IMHO key point, is much safer. I do suspect that failure rate when using TIM is much smaller then soldering so yield per 1k unit is probably much greater making it more $ efficient. Side effect here is increase in thermals, but within safe range, although more in upper spectrum, making it harder to OC using WC or AC solutions, so customers will most likely purchase next iteration sooner then it was before (due to smaller top clock speeds). Additionally it will force to de-lidding and personally I don't see many who would attempt it on 1k+ CPU, not to mention that switching to i.e. liquid metal require maintenance at least once per 1-2years and I personally wouldn't want to bother with whole process again. Of course this is My personal take on this, I'm not metallurgy expert and may be completely wrong why Intel does it. Also it seams like 14 ->18c SKU's will hit market in 2018 so AMD will have nice head start with their TR launch. PS. I was late to reply it seams.

It's poor greed and "I don't care they still buy it lolz" attitude. It's simple as that and there are no excuses. It's another example that capitalism without strong competition is not working very well for the consumers.

It's poor greed and "I don't care they still buy it lolz" attitude. It's simple as that and there are no excuses. It's another example that capitalism without strong competition is not working very well for the consumers.

Way to ignore everything posted previously addressing this in favor of just going 'wah wah'...:stewpid:

Everyone keeps saying "wait for benchmarks before deciding" but to me, I find both products to be ridiculous, and not in a good way. They get people to approach problems in a more lazy way - rather than try to optimize software to work the best way it can, people instead are like "meh, just throw more cores at it". All these products do is cater to people's impatience, egos, and wastefulness. If these were server chips, I wouldn't care - servers have a need for this many threads.

Unless Intel has purposely been pissing about and not working toward way better per core performance we're nearing a wall (at least for now). It was bound to happen eventually.

As for Threadripper having latency issues, people forget one thing: Each CCX in a 1800x is connected to the other via Infinity Fabric at the same speed as the memory controller. The exact same applies for connections to the rest of the CCX on a Threadripper CPU SoC. The "Numa" issue has already been addressed by the original Ryzen launch. As for this Intel platform, unless it cuts prices by at least 40% and even 60-80% for some models, it's DOA. I would argue it's DOA just by the artificial PCIe limitations.

@Chillin Thank You for Your educational post. Do You happen to have information about CPU failure rate due to solder degradation? This is actually crucial peace of information, that is probably key to this debate (of course from $ stand point) and how it compare to TIM solutions. Your points are valid, but a little bit off to me, i.e.: - shipping tolerance is standard shipping thermals that are used practically for any electrical device, i.e. 2nd gen cpus: intel.com/content/dam/www/public/us/en/documents/datasheets/2nd-gen-core-desktop-vol-1-datasheet.pdf So this is not something new and I do fail to see any correlation with argument TIM vs solder debate - I mean same spec for storage are used for Broadwell-E also: intel.com/content/www/us/en/processors/core/desktop-6th-gen-core-family-datasheet-vol-1.html - issue about TIM vs 3rd party solution is different and not really related to soldering issue, but yes currently used TIM is not bad as some may think Soldering and liquid metal are better thermal compound without a doubt, see this project for example: overclocking.guide/the-truth-about-cpu-soldering Difference on 6700K TIM vs solder is 18*C and liquid can do even better. Of course for average customer this is non issue (and average customer don't buy HEDT), simply Your CPU can't hit 4.4+GHz because it do overheat, so either spend another 100+USD for de-lidder + liquid and then bother with maintenance or purchase new faster CPU, but this is HEDT so some brave ppl will bother 😀. I do stand by my statement. Arguments like solder (probably) has higher failure rate is not really an issue as HEDT is for semi pro usage mostly and lifespan for that platform is around 5 years (I mean mostly now additional features, then actual CPU performance). If indeed there is higher failure rate, then still this is cost saving move (less RMA). Pro usage will go with Xeons (and they also have a TIM now) for maximum stability and reliability. Bottom line here, at least for me, is that Intel don't want customer to OC these CPU's and because TIM do save them more $$$ they go with it (on 3 fields - cheaper production, less RMA assuming superiority of TIM and no more high OC so next purchase is done faster by customer). I do understand that I may sound like opting for inferior / more failure prone solution but at this point I don't have any data to compare solder vs tim failing rates, if available please do correct me 🙂. Why do I bother about temps it's because I wan quiet system ideally passive (NSG S0 is coming in August so maybe such wish will come true, but dunno if price / noise ration will justify all of this) And it's not like desktop CPU's with TIM don't malfunction, I did read some time ago that one unlucky soul had to 3 times RMA 7700K (or 7600K) but it was probably not thermal related issue.

Unless Intel has purposely been pissing about and not working toward way better per core performance we're nearing a wall (at least for now). It was bound to happen eventually.

Yes and no. AMD's first attempt at Zen has already proved that it is possible to squeeze more performance out of x86. Future iterations are bound to be better. But even then, you are right that Intel (and AMD for that matter) are nearing a wall where there's not much left that can be done to improve performance. However, I was referring more to optimizing software. So many programs and drivers are so poorly written, or at least have a lot of room for improvement.

As for Threadripper having latency issues, people forget one thing: Each CCX in a 1800x is connected to the other via Infinity Fabric at the same speed as the memory controller. The exact same applies for connections to the rest of the CCX on a Threadripper CPU SoC. The "Numa" issue has already been addressed by the original Ryzen launch. As for this Intel platform, unless it cuts prices by at least 40% and even 60-80% for some models, it's DOA. I would argue it's DOA just by the artificial PCIe limitations.

ROFLMAO, yah it's totally dead /sarcasm off Sorry Cap'n Doom and Gloom, Intel (and the platform) will do just fine, as will AMD.

ROFLMAO, yah it's totally dead /sarcasm off Sorry Cap'n Doom and Gloom, Intel (and the platform) will do just fine, as will AMD.

Dude, even Intel-friendly and frequently Intel-sponsored mainstream YouTube channels like Linus' have issues with the new Intel platform. I don't understand how you can justify a $1000 10-core CPU which doesn't even seem to have a significant clock advantage at this point. Not to mention the lack of PCIe lanes and the rest. Not being able to see that in its current pricing at least, this is a bad product, it's just blind. I have an Intel CPU which is the best CPU I have ever purchased, period. I have no problem with Intel. The issue they have is that they have overdone it with the milking at this point, and they still behave as if they are the only CPU company around.

Dude, even Intel-friendly and frequently Intel-sponsored mainstream YouTube channels like Linus' have issues with the new Intel platform. I don't understand how you can justify a $1000 10-core CPU which doesn't even seem to have a significant clock advantage at this point. Not to mention the lack of PCIe lanes and the rest. Not being able to see that in its current pricing at least, this is a bad product, it's just blind. I have an Intel CPU which is the best CPU I have ever purchased, period. I have no problem with Intel. The issue they have is that they have overdone it with the milking at this point, and they still behave as if they are the only CPU company around.

Top 100 with 1 GPU http://www.3dmark.com/hall-of-fame-2/timespy+3dmark+score+performance+preset/version+1.0/1+gpu 2 GPUS http://www.3dmark.com/hall-of-fame-2/timespy+3dmark+score+performance+preset/version+1.0/2+gpu Not one AMD CPU F**k me, Intel is in serious trouble Yes, they are behaving like they are the only game in town because up until now, they have been. I've said it before in this thread, they will behave as though there is no competition, that's what Intel does. I'm not saying they aren't going to have to adjust their prices but saying the platform is DOA is silly and equally unsubstantiated since it hasn't even formerly released yet.

Sure, Intel is fine because the top overclocker in the world who even has his underwear sponsored has the top 3DMark score performed using a $1700 Intel CPU with LN2. What an amazing argument. :infinity:

Sure, Intel is fine because the top overclocker in the world who even has his underwear sponsored has the top 3DMark score performed using a $1700 Intel CPU with LN2. What an amazing argument. :infinity:

Yah, most of those (the top 100) are not LN2, speaking of 'amazing arguments'. Lets wait until we have numbers from X399 and X299, then we'll see what's what. And I tend to agree with Angel, short of TR being an unmitigated disaster, AMD will win this round, I just object to your assertion that X299 has failed when you have nothing to back up your claims.

Yah, most of those (the top 100) are not LN2, speaking of 'amazing arguments'. Lets wait until we have numbers from X399 and X299, then we'll see what's what. And I tend to agree with Angel, short of TR being an unmitigated disaster, AMD will win this round, I just object to your assertion that X299 has failed when you have nothing to back up your claims.

So, you are hired by Intel or just trying to justify your purchase? Tech doesn't have friends, if you mess up people will switch, that's how it works. If you don't like AMd, then great! but stop trying to defend your fanboy arguments... Intel LOST this paper launch, unless you can convince all their clients to pay 100% more for the same performance? not likely.