The Haswell Front End

Conroe was a very wide machine. It brought us the first 4-wide front end of any x86 micro-architecture, meaning it could fetch and decode up to 4 instructions in parallel. We've seen improvements to the front end since Conroe, but the overall machine width hasn't changed - even with Haswell.

Haswell leaves the overall pipeline untouched. It's still the same 14 - 19 stage pipeline that we saw with Sandy Bridge depending on whether or not the instruction is found in the uop cache (which happens around 80% of the time). L1/L2 cache latencies are unchanged as well. Since Nehalem, Intel's Core micro-architectures have supported execution of two instruction threads per core to improve execution hardware utilization. Haswell also supports 2-way SMT/Hyper Threading.

The front end remains 4-wide, although Haswell features a better branch predictor and hardware prefetcher so we'll see better efficiency. Since the pipeline depth hasn't increased but overall branch prediction accuracy is up we'll see a positive impact on overall IPC (instructions executed per clock). Haswell is also more aggressive on the speculative memory access side.

The image below is a crude representation I put together of the Haswell front end compared to the two previous tocks. If you click the buttons below you'll toggle between Haswell, Sandy Bridge and Nehalem diagrams, with major changes highlighted.


In short, there aren't many major, high-level changes to see here. Instructions are fetched at the top, sent through a bunch of steps before getting to the decoders where they're converted from macro-ops (x86 instructions) to an internally understood format known to Intel as micro-ops (or µops). The instruction fetcher can grab 4 - 5 x86 instructions at a time, and the decoders can output up to 4 micro-ops per clock.

Sandy Bridge introduced the 1.5K µop cache that caches decoded micro-ops. When future instruction fetch requests are made, if the instructions are contained within the µop cache everything north of the cache is powered down and the instructions are serviced from the µop cache. The decode stages are very power hungry so being able to skip them is a boon to power efficiency. There are also performance benefits as well. A hit in the µop cache reduces the effective integer pipeline to 14 stages, the same length as it was in Conroe in 2006. Haswell retains all of these benefits. Even the µop cache size remains unchanged at 1.5K micro-ops (approximately 6KB in size).

Although it's noted above as a new/changed block, the updated instruction decode queue (aka allocation queue) was actually one of the changes made to improve single threaded performance in Ivy Bridge.

The instruction decode queue (where instructions go after they've been decoded) is no longer statically partitioned between the two threads that each core can service.

The big changes in Haswell are at the back end of the pipeline, in the execution engine.

CPU Architecture Improvements: Background Prioritizing ILP
POST A COMMENT

248 Comments

View All Comments

  • dartox - Tuesday, November 27, 2012 - link

    Probably because most people know about how large an iPad is - if he said "tablet" form factor that's ambigious.. and if he said "Motorola XOOM" form factor not as many people are familiar with the size. Reply
  • kylewat - Tuesday, February 12, 2013 - link

    iPad is the correct nomenclature. Tablet is a word for things trying to be an iPad. Reply
  • Paer0 - Friday, October 05, 2012 - link

    Yes... Macs are well engineered and deliver a solid performance across board. Reply
  • stop-a - Saturday, October 06, 2012 - link

    100% agree on the well engineered part especially on the antenna gate when Steve God was saying "you're holding wrong", plus the recently ingeniously designed sapphire glass lens camera when Tim Schmok was saying "stay away from bright light source". Boy, Apple products must be engineered straight from the heaven; they are just too perfect for a mere earthling to use. Reply
  • Paer0 - Saturday, October 06, 2012 - link

    @stop-a. Since you are a 100% Apple hater, let me ask you this what computer do you use? And what OS do you use on it? I hope it doesn't crash several times a day. I use a MacBook Pro 2012 and I don't see anything come close. Reply
  • Urizane - Saturday, October 06, 2012 - link

    You really shouldn't use the 'crash several times a day' piece anymore. I'm annoyed every time I see this. My Windows 7 machine has an uptime of 20 days and counting. Most of the time, it waits for me to connect to it via SplashTop or FTP, or it's recording TV shows, but when I play games, I stress it bigtime. Seriously, stop with the Windows constantly crashes crap. It's just plain false now.

    P.S. - 20 days ago, I brought it to another house, thus the interruption in uptime.
    Reply
  • StevoLincolnite - Saturday, October 06, 2012 - link

    I have a Dual socket 2011 motherboard with dual Core i7 3930K's both chips clocked to 4.6ghz, 32gb of ram, Triple Radeon 7970 3gb cards powering 3x 27" Dell U2711 monitors in Eyefinity.

    Kay go. Lets see if your Mac can keep up or a Mac workstation at the same price. (Hint: Not going to happen.)

    Besides, Mac's look ugly, I prefer the whole she-bang of a side window with a nice water cooling loop and having the whole thing light up, not some dull silver box.

    Plus, my system is completely stable. Never had a crash yet with Windows 7 and... I have access to the last couple of decades worth of software and games, not to mention emulation of other platforms.

    I can also pretty much find software and hardware easily and it will "just work" I never have to ask the question of: Will it work on a Mac?
    Reply
  • lmcd - Saturday, October 06, 2012 - link

    I don't think you're in their target audience, for some reason. They're the best preconfigured system out there, especially once you ignore price. Reply
  • Magik_Breezy - Sunday, October 14, 2012 - link

    You'd hope a manufacturer can "configure" a system for an extra $1,400
    Hardware: $400
    I'm Apple: $1,400
    Total: $1,800

    With PCs manufacturers almost always loose money selling motherboards
    Reply
  • vt1hun - Tuesday, October 09, 2012 - link

    Two Core i7s working on a dual 2011 socket motherboard? You need QPI links for that which only certain Xeons have. Sounds like your system will just NOT work ! Reply

Log in

Don't have an account? Sign up now