Jump to content

PeterBocan

Member
  • Posts

    201
  • Joined

  • Last visited

Posts posted by PeterBocan

  1. 6 minutes ago, Dracarris said:

    If they can beat 98% of price-equivalent products, that would actually be a very big achievement - given that usually you are charged the hefty Apple tax compared to competitor models with equivalent performance.

    I don't trust that number. How do they know who has what? What are the Intel and AMD sales? I really don't think that the 98% is accurate.

  2. Just now, NotTheFirstDaniel said:

    Yeah my bad, I should've clarified America. But remember the gross salary isn't what we actually take home. A percentage is taxed, then another percentage goes to Social Security, then some goes to private healthcare (oh yeah because in the 21st century the best we have is the ACA).

     

    we have that here as well, it's just average and median salaries ale somewhere completely different.

  3. 1 minute ago, NotTheFirstDaniel said:

    Single Core? Multi Core is slammed by the i9 and AMD since you're comparing a quad-core to an octacore, but it's not like Geekbench combines the might of all 4 cores into one for ARM.

     

    And I think the low-power cores are the replacement for hyperthreading.

    yeah exactly, Octa*-core. It's not really octa core, it's more of a 4+4 so on paper it's better than 4 core and worse than full 8 core.

  4. 9 minutes ago, NotTheFirstDaniel said:

    Am I missing something? Did I watch a different keynote? Was I just absorbed by Apple's marketing (most likely answer)?

     

    The M1 chip is an ARM SOC based off of an Apple-designed core in a big.Little configuration. Per Apple's words, the M1 SOC is split into 4 high performance cores, and 4 high efficiency cores. They said the high-efficiency cores match the Core i3 in the MacBook Air, or the Core i3-1000NG4, an adaptation of an Ice Lake core. The difference being the M1 does it at significantly less power. Like significantly less power.

     

    Assuming the M1 chip is the same chip as found in the "leaked benchmark", it has a single core score of 1634. That score is 200 points higher than the Core i9-10900k. Now, people say it's apples and oranges, I don't know how it is but that's not my expertise I can't really talk, but taken at face value the M1 chip is very powerful while using a fraction of the power the i9 mobile chip uses.

     

    We will see next week whether I'm right or wrong, but if I'm right you should have nothing to worry about. If they can pull off outstanding performance on a 15-20W chip, then what the fuck can the do on a 45W chip? I'm excited. I really think they could match or maybe even slightly beat a Zen3 Threadripper based off of this. Hopefully by then though it doesn't take up 500W of power. They overbuilt the Mac Pro cooling system for some reason, and I think this is the reason.

    I presume those are benchmarks of the high-perf cores. So you are missing whole 4 high-perf cores of i9-10900k.Also wondering about "Hyper-threading".

  5. 1 minute ago, NotTheFirstDaniel said:
    Just now, Vitamanic said:

    Kind of odd that the new MacBook Pro 13 has halved memory and storage support.

     

    New model has a max of 16GB of RAM and 2TB of storage vs the 32GB and 4TB available on the Intel models.

     

    Also curious as to how well the ARM chips will stack up in the 16 inch model... If it takes 8 cores to match a 2 core 15w i3 chip, how many cores will it take to match my 45w i9-9980HK?!?! 
     

    I’m getting the feeling that they’re struggling to match 35-45w mobile chips. I can’t imagine them squeezing out 10x more performance to compete. Hopefully I’m wrong.

     

    that's because that black tar chip does not have the same cooling capabilities as open-die Intel/AMD. They just can't push it that far.

  6. 2 minutes ago, NotTheFirstDaniel said:

    How would you measure the speed of a CPU? If you measure just instruction speed, the M1 will always be faster, not accounting for the hardware-decoders + the times where multiple instructions are needed to simulate an x86 one.

     

    But I think it's still a net gain of speed, and native ARM code is going to fly with it. That's just me though I'd like someone to explain it indepth.

    in general you have IPC and you have clock. But then you have caches, you have software optimizations, you have SIMD, you have branch predictors. It all plays a big role in performance. It's all very relative so you need to fix some variables down, i.e. software and software optimizations.

  7. 1 minute ago, GoodBytes said:

    Actually the system should perform fine.

    It uses the latest just released from ARM, Cortex CPU architecture, which bring the CPU, not only faster, but also Octo Core.

    Already the Surface Pro X demonstrated with Qualcomm chip that the device is very competent and provide a great (performance wise) experience. Of course, its issue is that the CPU lacks performance for any x86 demanding applications. But this new CPU should allow Apple to deliver a good experience, and I expect, the day, Qualcomm decides to use it, that the latest Surface Pro X can deliver more interesting performance on x86 side (and x86-64 translation support should be coming next year)

    Octo*-core. It's 4 perf and 4 low-perf cores. It's not comparable to Intel's offering, though it's 14 nm vs 5.

  8. I came here to vent my frustration. Apple can not offer 8 and/or 16 GB RAM and call it a "pro" - how am I supposed to run Chrome with my VMs? :D
    Sure they swapped only the "base" models of Macbooks but unless they have a M1X chip with 32GB (which I assume they don't) it won't be usable. My assumption is that they sacrifice bandwidth on the M.2 SSDs to drastically swap unnecessary stuff. Also 1500$ for 8GB model of Macbook Pro model? Who is going to buy that? You can't offer "up to 32GB" one year and the next one "up to 16GB". I am kinda torn between buying what is still available before and they completely cut off Intel lineup.   
     

  9. Hey guys, I am considering buying two monitors, I am currently deciding between

    Dell Ultrasharp U2719D 27" 1440p currently on sale for 261 GBP (price lowered from 371 GBP)

    Dell S2721QS 27" 4K currently on sale for 250 GBP (price lowered from 317 GBP)


    I have been a previous user of Ultrasharp display and I was amazed by the quality of the display and thus I am a bit biased towards the Ultrasharp line, however they are on a bit pricey side and spec-wise under-performing. I am currently rocking Asus PB278Q 4K, old and heavy monitor (but it still works!). If you have any other recommendations for 2 monitors, for around 500 please let me know. I want to Vesa mount the monitors. Thanks!

  10. 1 minute ago, rcmaehl said:

    Do we REALLY need a smaller nvme drive setups? I recently worked on my younger sister's Dell G3. The internal 256GB m.2 SSD was the size of a dime. 

     

    At this point we're just asking to lose full size SSDs as if they were micro SD cards.

     

    I am taking a stand right now that I should not be able to lose an SSD on the floor or in a couch.

    it may come in handy. Tablets? Cell Phones? IoT devices?

  11. 2 minutes ago, captain_to_fire said:

    Just so you know, Apple isn't the only one soldering SSDs to the motherboard.

    Yes, but I own Apple product and I could not say the other brand names.

     

    5 minutes ago, Mira Yurizaki said:

    It may be good to have something like this, but remember:

    standards.png

    Agreed, but there's a need for a little advancement.

  12. Quote

    At the Flash Memory Summit today, Toshiba introduced a new form factor for NVMe SSDs that is small enough to be a removable alternative to soldered-down BGA SSDs. "The new XFMEXPRESS form factor allows for two or four PCIe lanes while taking up much less space than even the smallest M.2 22x30mm card size," reports AnandTech. "The XFMEXPRESS card size is 18x14x1.4mm, slightly larger and thicker than a microSD card. It mounts into a latching socket that increases the footprint up to 22.2x17.75x2.2mm." 

     

    Quote

    XFMEXPRESS is intended to bring the benefits of replaceable storage to devices that would normally be stuck with soldered BGA SSDs or eMMC and UFS modules. For consumer devices this opens the way for aftermarket capacity upgrades, and for embedded devices that need to be serviceable this can permit smaller overall dimensions. Device manufacturers also get a bit of supply chain flexibility since storage capacity can be adjusted later in the assembly process. XFMEXPRESS is not intended to be used as an externally-accessible slot like SD cards; swapping out an XFMEXPRESS SSD will require opening up the case of the device it's installed in, though unlike M.2 SSDs the XFMEXPRESS socket and retention mechanism itself is tool-less. 

     

    I am actually pretty stoked about this news as some notebook vendors, *cough*Apple*cough*, could stop soldering their SSDs into Motherboards.

     

    https://hardware.slashdot.org/story/19/08/06/1911212/toshiba-introduces-new-tiny-nvme-ssd-form-factor

×