Intel's 'Haswell' CPU refresh expected early next month alongside 9-series chipset

[Shawn Knight]

Intel’s next generation Haswell processors could hit the market as early as next month, well ahead of Computex 2014. Led by the Core i7-4790K, the chips could appear on May 10 in most markets according to sources in IT retail as reported by techPowerUp.

The refresh is said to offer marginally better performance compared to current generation Haswell chips and will displace existing chips from their current price points. The 9-series chipset, meanwhile, includes features such as M.2 solid state drive support which the publication tells us will get everyone ready for the flood of 1,000MB/s solid state drives expected to launch around Computex.

Intel was originally expected to launch the Haswell refresh in April and distribute the chipsets in May but based on feedback from motherboard clients and channel retail partners, they elected to postpone the CPU launch and release the two products around the same time in early May.

New socket LGA1150 boards based on Intel’s Z97 Express chipset are expected to arrive in mass in the week leading up to the CPU launch.

I’ve been eyeballing a system upgrade for a few months now to replace my aging Core i5 Sandy Bridge CPU so this may be the perfect opportunity to do so. What’s more, the lure of 1,000 MB/sec solid state drives is hard to ignore.

Are you planning to upgrade to a new Haswell CPU?

Permalink to story.

https://www.techspot.com/news/56346-intels-haswell-cpu-refresh-expected-early-next-month-alongside-9-series-chipset.html

 

[TomSEA]

"Are you planning to upgrade to a new Haswell CPU?"

Nope, waiting on the Broadwell chips out later this year, early next.

 

[GhostRyder]

I'm waiting on the Haswell-E series and the X99 Chipset to come out so I can grab one of those 8-Core chips at the 500 dollar price point.

 

[Matt W.]

Current rig:

Asus P5K-E BIOS: 1305
Intel CoD E8400 @ 3Ghz
6Gb of DDR2-800 (1Gb x 2, 2Gb x 2)
eVGA nVidia 8800 (640Mb)
320Gb SATA I, 500Gb SATA II

With no intent on purchasing anything soon, I would be holding out for Broadwell at a minimum. I've been eyeing the Samsung EVO 512Gb SSD now at some $260, but life is just to tempting to store everything (which doesn't accumulate to more than 600Gb for the most part) on a drive which can pull 1Gb transfers on a more advanced chipset. Primarily the startup of Steam and it's apps for this user. Otherwise, I am in need of a simple Nexus 7" (2013). There's just to much depreciation in the value of hardware. After building this DX10 system here in the start of '07 and looking back, there is just to much hardcore stuff I intend to support - this given, high PPI 4K displays making the mainstream and nVidia Pascal tech' being on its way. Along with it all, PCIe 3, and 4 down the road somewhere to consider and DDR4. We're just "at that point" for this consumer.

 

[VitalyT]

For those who use an overclocked i7-4770K, like myself, there is no reason to look at i7-4790K, as it is almost identical.

It is also doubtful what Intel insinuates about M2 adaptation in desktops, implying mass production of those. First of all, the performance increase isn't that big, and second - production of mobile SSD-s for desktops in itself sounds a bit stupid - does it not?

 

[Guest]

So they havn't heard of
Z drive or revodrive 3
both of these can easily achieve 1.5 Mps and have done for a few years now..

 

[Guest]

So they havn't heard of
Z drive or revodrive 3
both of these can easily achieve 1.5 Mps and have done for a few years now..

correction 1.5KMps ^^

 

[wastedkill]

Arent the 16core extreme edition chips coming out this year or early next year?

 

[GhostRyder]

Arent the 16core extreme edition chips coming out this year or early next year?

There are some 8 Core 16 thread Haswell-E chips coming out this year yes.

 

[theBest11778]

Meh... My Core i5 2500K's fine for everything out right now, and the foreseeable future. Waiting for DDR4, PCIe 4, Sata Express, HDMI 2.0 (On GPU,) and >$500 32-37" 4K/UHD 60hz Monitors/TVs before I consider upgrading.

 

[hahahanoobs]

Also waiting for Broadwell-E and X99 chipset. I could also use a high refresh rate monitor. GPU is still up in the air, but I'm in no rush right now.

 

[hahahanoobs]

Arent the 16core extreme edition chips coming out this year or early next year?

No. Native 15 core for server and 8 core for desktop is all we get.

 

[Holotus]

Everytime these chip companies reveal something new I completely loose interest. You really have to wait every 3-4 years to see a significant jump in power. Yearly releases is just minor tweaks to raise the numbers to make it seem better.

 

[Raoul Duke]

For once I'm more excited about an improved chipset than the cpu's.

 

[Darth Shiv]

So they havn't heard of
Z drive or revodrive 3
both of these can easily achieve 1.5 Mps and have done for a few years now..

Those drives didn't have TRIM support I believe and were also exorbitantly priced. These drives are more consumer level pricing and therefore adoption.

 

[Mr Majestyk]

For those who use an overclocked i7-4770K, like myself, there is no reason to look at i7-4790K, as it is almost identical.

It is also doubtful what Intel insinuates about M2 adaptation in desktops, implying mass production of those. First of all, the performance increase isn't that big, and second - production of mobile SSD-s for desktops in itself sounds a bit stupid - does it not?

Well the new K series get a better quality thermal paste under the heat spreader which should allow them to OC much better than Ivy Bridge or current Haswell and hopefully more like the Sandybridge CPU's which had a high quality tantalum based thermal paste.

 

[TheLastPanda]

Meh... My Core i5 2500K's fine for everything out right now, and the foreseeable future. Waiting for DDR4, PCIe 4, Sata Express, HDMI 2.0 (On GPU,) and >$500 32-37" 4K/UHD 60hz Monitors/TVs before I consider upgrading.

Out of curiosity, why is anyone waiting for pcie4? I have a second gen pcie slot running at x8, and it benchmarks the same in terms of gpu performance as pcie3 x16. There just isn't a bottleneck that I can see for quit a ways down the road. Is there something I don't know about, or is everyone planning on pcie solid States exploding soon?

 

[DDG4005]

I'm running two boxes with Core i7 3770s on Z77 motherboards (Asus's P8Z77-V Premium) so there's no reason for me to upgrade at this time. Still, it's good to see new tech arriving for the summer; it reminds me of my upgrade to a Pentium II 350MHz CPU and a 440BX-powered motherboard from a Pentium 233MHz processor and 430TX-based board back in June of '98 (just a week and half shy of Windows 98's release). Good times .

 

[theBest11778]

Out of curiosity, why is anyone waiting for pcie4? I have a second gen pcie slot running at x8, and it benchmarks the same in terms of gpu performance as pcie3 x16. There just isn't a bottleneck that I can see for quit a ways down the road. Is there something I don't know about, or is everyone planning on pcie solid States exploding soon?

When the GTX 680 came out TechPower up did an article about the differences in PCIe performance. In short a PCIe 1.0 16x slot, PCIe 2.0 8x slot, and PCIe 3.0 4x slot were all capable of providing perfect bandwidth. With the new Maxwell GPUs looking to at least double the GTX 680 in performance it would make sense those numbers will also double. SO PCIe 1.0 is out of the question, 2.0 will require 16x, and PCIe 3.0 will require an 8x slot for optimal performance. Now with other things like SATA Express, and other devices starting to leech on the PCIe bus there will be fewer dedicated lanes for just GPUs. I haven't heard Intel say anything about increasing the 16 lanes on their CPU die (Sure the E series have more, but most people don't go that route.) So waiting for PCIe 4.0 just means that those same 16 lanes can feed any new upgrades that will be added and still have some headroom. PCIe 3.0 should work, but might be pushed to it's limits.

 

[theBest11778]

Well the new K series get a better quality thermal paste under the heat spreader which should allow them to OC much better than Ivy Bridge or current Haswell and hopefully more like the Sandybridge CPU's which had a high quality tantalum based thermal paste.

Sandy Bridge CPUs were just soldered together, I believe you're right that these new Haswells will be soldered as well (Remember reading it somewhere.) I could never get an Ivy Bridge 3570K or Haswell 4670K above 4.2Ghz stable. They were quite disappointing. I switched back to a 2500K just to keep load temps under control.

 

[TheLastPanda]

In short a PCIe 1.0 16x slot, PCIe 2.0 8x slot, and PCIe 3.0 4x slot were all capable of providing perfect bandwidth. With the new Maxwell GPUs looking to at least double the GTX 680 in performance it would make sense those numbers will also double

I am not sure I agree with this. I see what you are saying, but if we did not run in to a bottleneck using a 680 in PCIe 1.0x 16 then it is possible the bottleneck is further off then we think. Maybe it would take a card three times as bandwidth intensive to hit that wall, or maybe it is right there. The point is you can't assume anything about the bottleneck from not finding one. I see the utility for it but I don't see why people seem to get caught up on thinking it is better because it alleviates the bottleneck X component wont let it get to more so then we talk about finding what X component is.

Also, I don't want to sound nit picky, but "At least double" seems off base to me. From Nvidias own site: http://international.download.nvidia.com/webassets/en_US/shared/images/products/shared/lineup.png

This shows the titan black performing about 25% better. assuming the 680 is at 4100 and the titan is at 5200 which is my closest guess. A card at least double in the same benchmark would be over 8000 which is a mid 50's% jump over the flagship single gpu (not even considered to be a consumer level product). You said "GPUs" as well implying that not just the one on the very top of the 800 line will manage this. Even assuming that does happen plus all of the following: the bottleneck is currently right at pcie3 x4, bandwidth scales linearly with performance, cpu's will support more lanes on that socket eventually before an upgrade is neeed, and that more lanes will be needed for new devices..... then why is pcie3 x16 not good enough for consumer grade computing? Its a lot of assumptions to make to get a few fps from bleeding edge hardware that may need it years down the road....by the time Nvidias new socket may be in use already.

 

[Lionvibez]

So they havn't heard of
Z drive or revodrive 3
both of these can easily achieve 1.5 Mps and have done for a few years now..

You can keep your OCZ drive.

I will take a drive that is validated,tested and branded by intel even if its alittle slower.

In my system Stability of the storage system is king over sequential transfer speeds which everyone seems to drool over.

 

[dividebyzero]

I am not sure I agree with this. I see what you are saying, but if we did not run in to a bottleneck using a 680 in PCIe 1.0x 16 then it is possible the bottleneck is further off then we think.

If you're talking about GPUs, and particularly gaming, internal bandwidth (with a preference for bus width over frequency) is more important than system bus (PCI-E) bandwidth. Games and consumer apps that rely upon the GPU typically only use the CPU for A.I routines, CPU physics, driver draw calls, app/GPU resource allocation, and shader compilation predominantly, and the first of those is increasingly becoming a compute shader (GPU) operation. The AI component being thread and bandwidth sensitive is also the primary reason why RTS games tend to be much more dependent upon CPU thread/clockspeed. As a general rule, the CPU<-> System RAM tend to stall before PCI-E bandwidth becomes an issue.

Anyhow, here's that TPU PCIE scaling article, and a similar Titan SLI article from Anand - note the difference between in scaling between Crysis 3 (GPU orientated), and Shogun 2 ( CPU A.I. intensive)

Maybe it would take a card three times as bandwidth intensive to hit that wall, or maybe it is right there

PCI-E bandwidth can easily be saturated by GPUs - note the plural (or storage controllers and adapters and the like for servers/data centers/super's), just not consumer apps unless you really load up the system bus.

Also, I don't want to sound nit picky, but "At least double" seems off base to me.... You said "GPUs" as well implying that not just the one on the very top of the 800 line will manage this. Even assuming that does happen plus all of the following: the bottleneck is currently right at pcie3 x4, bandwidth scales linearly with performance, cpu's will support more lanes on that socket eventually before an upgrade is neeed, and that more lanes will be needed for new devices..... then why is pcie3 x16 not good enough for consumer grade computing?

The whole GPU power needs external bandwidth paradigm is probably due to die a death over the next few years. Nvidia and AMD both seem intent on incorporating ARM cores into the GPU package, which should -amongst other things, decrease dependency on the system CPU. It is why both vendors are looking to greatly increase internal bandwidth ( HBM, WideIO2 etc). If PCI-E 3.0 and 4.0 ever become valid "must have" features it will likely be because of storage hardware advances.

 

[Blue Falcon]

Waiting to see if Intel offers a $500-600 8-core Haswell-E. I think Intel will keep the 6-core at $500 and use the 8-core at the $999 level which is good differentiator for enthusiasts. In that case, I'll wait until 20nm Maxwell big die / 20nm Pirate Islands flagship and upgrade everything at once on Skylake. I will not go beyond 2 GPUs which means PCIe 3.0 x8/x8 should be more than sufficient for my needs. I have a feeling the Haswell-E problem will face high DDR4 prices though.

 

[GhostRyder]

Waiting to see if Intel offers a $500-600 8-core Haswell-E. I think Intel will keep the 6-core at $500 and use the 8-core at the $999 level which is good differentiator for enthusiasts. In that case, I'll wait until 20nm Maxwell big die / 20nm Pirate Islands flagship and upgrade everything at once on Skylake. I will not go beyond 2 GPUs which means PCIe 3.0 x8/x8 should be more than sufficient for my needs. I have a feeling the Haswell-E problem will face high DDR4 prices though.

Oh wow thanks for jinxing things, I hope to heck they don't do that

I was more thinking they might do something like this:

Drop the Quad core completely from the lineup, make the 6 core the low end one (Probably still price higher than the old quad was at 400 maybe), have a 8 core K model at 5-600, then have the usual 1k edition 8 core.


I was personally hoping to grab one of those 5-600 dollar models soon hoping it was going to be an 8 core edition. If they do what you said and make the 1k models be the only 8 core editions I would be pissed at that price lol.