Intel screws high end users yet again.

[Buckeye]

New Tech is coming out so fast these days that its a waste of money and resouces to keep out dated platforms up to date it seems.

Socket 1366 boards are getting hard to find these days and tbh 920's, 990x's are still plenty good CPU's to run. I had my old trusty Giga UD3 finnaly go out on me after 3 years or so which I was running a 990x @ 4.8ghz and used my Giga UD7 and 2500K @ 5.0ghz as a replacement until I get a 3960x system up and running, I feel like I took a huge step backwards. Don't get me worng tho the 2500K system does just fine but I have gotten used to more cores & memory.

Just watch as the people who end up getting ARM or ATOM laptops and tablets start using them, loading up there junk and start complaining about how slow they are. ARM & ATOMs suck and we all know it, but people buy them to save a few bucks then complain about performance.

As far as TRIM goes I don't think I have run a platform that has ever used it as I mostly run Raid 0's all the way back to my older MTRON Battleship which lasted 5 years, now replaced by two 120gb Intel 520's and I get better performance vs 8 of the MTRONs. Its not the end of the world to me at this time because it won't be that long until I upgrade to a platform that I can use TRIM on Raid's.

Why don't we see 8 core desktop CPU's ? Main reason is heat, and well Intel is stuck with the mindset that they know best. We all know that we can use a nice water cooling setup to cool those down but they have to design them to run with air coolers or they would get to many blown up CPU's in the RMA channel.

The server world is waking up also for some applications. I build high speed servers for a living and at first my clients don't believe that a server can run at 5.0ghz so they get a test unit. First thing they say is OMG !! then we get phone calls like... I just keep throwing stuff at it and it doesn't slow down. We have our HSSM - High Speed Server Modules installed at Equinox NY4, Chicago, LA and this month in the UK and every single client says they will trade stuff like IPMI over speed anyday as there are ways to work around it. Will we ever see unlocked Xeon E5-2687W's, I don't think so which is a shame as I would kill for those, so would my clients.

[lutjens]

It's interesting to hear that servers are overclocked on a regular basis and that people are happy with the results and the performance. What aggravates me is that because the chips are locked, we don't even have the option. If they were unlocked, we'd have to option to bump up the multiplier if we wanted (and indeed $2000 should get us that ability), or simply leave the chip at default. A server from any of the OEMs would be completely unaffected, but for those who have the right board and the expertise, it would be incredible. There are many professional segments like day traders, graphics artists and the like that would likely jump at the opportunity to squeeze out a few more MHz from these chips and would ensure the environment in which they run is suitable for the operation at higher speeds. For the rest of us who like to tinker, dual unlocked E5-2687Ws under water at 4.5GHz+ would provide for an incredibly powerful system and Intel would sell many more chips as a result.

[Ket]

But why should you have to? Intel's too lazy and cheap to do the right thing and support the users of their older products, even when one of those older products is their "high end" tier and has been around for much less than one year . It shows the complete lack of interest that Intel has in keeping customers happy at this point. They've attached support for RAID 0 TRIM only to their newest product as a way to force people to upgrade. It shows exactly how much Intel stands behind its products...after they have your money, they don't care. I'm very surprised they haven't cut their warranty on retail CPUs down hard yet (although I'm sure that's coming)

I'm not saying what intel is doing is right, just that people like me are the resistance. We will fight on.

[Stevethegreat]

I'm sure it won't be impossible to add raid-0 trim support on X58 chipsets, if not officially certainly through the unofficial route.

Personally -4 years now- I found no reason to migrate from my x58 setup. It's not that I'm being cheap, it's only that the new chips are not clearly better, in fact they may even be less capable. I have triple channel, unlocked 6-core support, 52 pci express 2.0 lanes (granted it's not 3.0 but it holds up nicely). I'm not going to lose all that merely to get raid-0 trim support (even though I'm on raid0 for years now as well)... The unofficial route is the way to go, don't care for the new chipsets unless they start being as awesome as they used to be...

[Zaskar]

I dont get it, thats the standard RST driver, doesn't all X79 boards use the RSTe drivers which are on a separate development branch that is usually a bit behind the normal drivers? (enterprise drivers require more testing to be considered stable enough)

Hell, if your not using a beta RSTe driver from a 3rd party site like Station-Drivers then your not even getting TRIM on individual SSD's if the controller is set to Raid. That now standard RST feature hasnt yet trickled down to the stable RSTe driver yet so its probably going to be a while before we get this new feature on the enterprise branch.

[SKYMTL]

Not so fast guys:

Intel’s TRIM Support for RAID Coming to X79 "Soon"

[masterg]

somebody with weight yell at them?

[Zaxx]

Only P67, Z68, Z77, H67, H77, Q67 and Q77 chipsets are currently supported, which tends to leave some Intel’s highest paying clients (enthusiasts and power users) out in the cold so to speak. Older motherboards based on chipsets not mentioned above aren’t supported either, but that’s likely due to the complications and limited ROI involved with porting the newest driver structures over to EOL’d platforms.

Does this mean series 6 and 7 will both end up with trim for raid0?

[Nikolasz]

Amd got 16 cores.

http://www.amd.com/us/products/serve...l-numbers.aspx
http://www.amd.com/us/products/prici...r-opteron.aspx

And what you guys need to do, is get one cpu, mainboard

[SKYMTL]

Does this mean series 6 and 7 will both end up with trim for raid0?

Looks like it. If you look at the release notes, several of Intel's 6 series boards are listed as supported.

[naokaji]

Does this mean series 6 and 7 will both end up with trim for raid0?

In theory, but only if the mainboard manufacturer can be bothered to releases a bios update as the driver is just one of the requirements.

[Zaxx]

Updating the orom in the bios isn't a problem...

[DefStar]

Is this a conspiracy theory forum?

[Vicey]

Sure is a lot of assumptions going on here. Intel has only just launched this feature. You need to give them at least 1-2 Months before you can conclude they won't be delivering this feature to X79 chipsets in my opinion.

[rintamarotta]

I have no problem modifying my Z68 mb bios's (MIVE-Z) to make Intel RST driver detect my mb's chipset as Z77, i have allready modified my bios with latest Intel Raid Rom.

If it works on Z77, it will work on Z68, eventually.

[SKYMTL]

Sure is a lot of assumptions going on here. Intel has only just launched this feature. You need to give them at least 1-2 Months before you can conclude they won't be delivering this feature to X79 chipsets in my opinion.

Our news post already addressed this. Intel has stated publicly that an RSTe driver for X79 is incoming.

[lutjens]

Intel’s feedback also brings up a good point: the RST development for the X79 chipset is on the RSTe (or Enterprise) branch rather than the standard RST development cycle used for other mass market chipsets. This process with its extra validation steps is likely the main reason why we have not seen an RSTe driver that supports RAID 0 yet.

Thanks for the update SKYMTL. The explanation around RST and RSTe branches makes sense regarding X79 support and is greatly appreciated. Clarification regarding upcoming support should have been released by Intel in the release notes when the driver was released to avoid confusion, but its nice to know that support will eventually materialize. While this is indeed good news, it's unfortunate that the X58 chipset isn't being supported as well, even on a basic RST (non-e) level of validation. The driver should work on the X58 (not be intentionally blocked), even if Intel made it available on an "at your own risk" basis. While still quite robust, the platform is more than 3 years old and I understand why they wouldn't want to bother with it. It's not near the outrage that the X79 promised to be.

[Kain665]

Lutjens. Stop being angry and go outside.

[lutjens]

Lutjens. Stop being angry and go outside.

Sorry, I get annoyed when a company I've supported very well over the years tries to screw me, and I'm not afraid to voice my displeasure. I'm not willing to simply take what I'm given if its insufficient. The consumer's position in the market would be much better off if there were more people who actually stood up and made their voice heard instead of rolling over and mutely taking what they're given with no complaint.

[Kain665]

Vote with your money, if you care so much.

[generics_user]

i really hope for P67 support; i'm at 95% capacity on my 128gb SSD 830 and adding a second drive is cheaper than buying a 256gb drive

[Andrew LB]

somebody with weight yell at them?

I think its more likely that someone with a bone to pick with intel started making a fuss as commonly seen on the internet, so Mr. Anand decided to validate or disprove rumor mill claims. Smart journalists will typically refrain from any serious criticism prior to the company making an official statement, a practice that doesn't seem to be very common these days.

[sholvaco]

Vote with your money, if you care so much.

You know I see this phrase thrown around a lot, but I can assure you from personal experience that voting with your wallet has about as much effect against corporate practices as a torn condom has against unwanted pregnancy.

One can abstain from buying products which:
-have a major defect in it's supporting platform due to gross negligence (which had been deliberately kept secret until after launch in order to meet fiscal projections),
-have most of the promised new platform features disabled in firmware due to inability to deliver sufficient stability within fiscal time frame (gotta launch that badboy anyway),
-have their performance potential deliberately limited to protect another market segment,
but for every well informed individual that chooses to do so there are hundreds who sheepishly stand in line waving their money at Intel's latest and "greatest" toys.

I've sold my LGA775 build in January of 2011 eagerly anticipating Sandy Bridge, yet due to the factors mentioned above I have yet to own a new system of similar capability. I've been voting with my wallet ever since and yet Intel's commitment to deliver has only gotten worse in the meantime.

On the other hand the Internet gives potential costumers like us a wonderful ability to voice our opinions and for these opinions to actually be heard. Raising a stink against a company on a public forum can have an actual effect I've found out, as many of my personal complaints or suggestions were somehow fixed/realized within successive product line ups.

When somebody presents an argumented complaint that you disagree with, please refrain from attacking them with cheap shots like accusations of conspiracism (defstar) or platitudes like "vote with your money". Either present an opposing opinion of your own or set them straight with legitimately sourced facts if you think they're completely wrong.

You may be satisfied with Intel's current offerings and have no issue with their corporate practices but lutjens clearly thinks otherwise and so do I... and I'm glad he's the one raising these issues because frankly I've gotten too lazy and quite simply disillusioned to the point where i can barely muster a fck.

[Circaflex]

http://www.techpowerup.com/170822/TR...ming-Soon.html

While the PC enthusiast community at large welcomed news of SSD TRIM command being made possible in RAID 0 setups, provided Intel 7-series chipset RAID controllers are used, users of Sandy Bridge-E HEDT platform, running Intel X79 chipset, didn't. The feature was advertised to be initially available only to users of 7-series client chipset (such as Z77, H77), with Rapid Storage Technology (RST) driver version 11.0 and above. Intel made amends by announcing that users of Intel X79 chipset will be able to take advantage of TRIM command in RAID 0 setups, with a future version of RSTe (Rapid Storage Technology Enterprise) driver, without being specific about the version or its release date.

Said Intel in its note, "Current RSTe drivers specific to X79 do NOT support TRIM on RAID 0, but an updated RST driver version coming soon will add support for X79 based systems, including the TRIM on RAID 0 feature. Note that on client 7-series chipsets (non X79), RST driver version 11.0 and beyond supports TRIM on RAID 0."

[dartaz]

Amd got 16 cores.

http://www.amd.com/us/products/serve...l-numbers.aspx
http://www.amd.com/us/products/prici...r-opteron.aspx

And what you guys need to do, is get one cpu, mainboard

16 core ! So what ?

1-Those things have very low clock speed.

2-Those things are based on crappydozer which have very low IPC.

Low clock speed + low IPC = horrible single-threaded performance

Even in heavy multi-threaded benchmarks it doesn't perform that good. Do not let those 16 cores fool you

Here check a comparison between Xeon X5670 (6-core Gulftown) vs Opteron 6276 (16 core crappydozer) in heavy multi-threaded such as cinebench and 7-zip
http://images.anandtech.com/graphs/graph5058/42233.png
http://images.anandtech.com/graphs/graph5058/42258.png


Opteron 6276 costs $900 in newegg. I don't see any reason to buy this over i7 3930K specially if you not building a server machine. Opteron 6276 is going to be day and night slower in gaming and lightly-threaded application. And in multi-threaded app it won't beat 3930K