Gigabyte BRIX S Broadwell Ultra Compact PC

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
Those of you entertaining the idea of buying an ultra compact computer should head on over to PC Perspective this afternoon and read their review of the Gigabyte BRIX S BXi7H-5500 Broadwell.

The BRIX S that I tested in this review has a lot to offer as well. The performance is fantastic, as I discussed above, and it should be able to address nearly any usage model save for video editing and gaming. Intel's new Broadwell architecture is responsible for a lot of that but Gigabyte deserves credit for including the key input and output features like USB 3.0 and HDMI/DisplayPort.
 
GPU performance on that little unit is incredible! Similar speeds to a GeForce GT 540M
 
Holy crap, better than an A10-5800K? That's some serious power there! I can't wait to see how it would perform under actual gaming load.
 
I can't wait to see actual HD6000 and Iris PRO 6100 - the last one should totally own integrated graphics.
 
I can't wait to see actual HD6000 and Iris PRO 6100 - the last one should totally own integrated graphics.

Yeah the desktop processors with Iris 6100 will be amazing for just-mITX builds. I wonder whether Thin-mITX builds with that will be possible or if the CPUs with Iris are going to have a TDP too high for those boards.
 
Yeah, but what I'm waiting for most are the Pro versions that have dedicated cache memory. hope the brix will have those too.
 
I can't wait to see actual HD6000 and Iris PRO 6100 - the last one should totally own integrated graphics.

2015 is the year of the APU wars between AMD and Intel. It's also the beginning of the death of the $100 graphics card market. Just wait for Intel's Skylake. If AMD doesn't surpass Intel in graphics then they're SOL. I doubt it, so they'll have to go beyond that of Intel's new graphics.

You could build a superior gaming PC compared to the PS4 and Xbone without the need to buy a graphics card. Why you think Valve is pushing Intel to fix their driver performance in Linux? Intel graphics aren't going to be a joke for much longer.
 
Yeah I get it, I just wish intel went for 128 shading units rather adding just few % more each iteration. Like make a gaming chip family which is 10W more power hungry and put those 25W units instead of 15W ones inside NUCs.

As for the intel's linux drivers - those as I remember were the best linux drivers like forever, valve didn't need to push anything. nvidia had good proprietary ones and amd was like wtf i need to do to even watch a movie or webpage without tearing.
 
Seems like a hell of a lot for the size

That being said, the name "brix" always amused me.
 
over $500 though...

Too bad. Would have been a nice way to expand with more MythTV frontends if it were cheaper...
 
If they were available and reasonably priced, I would totally want an ultrabook with HD 6000.
 
while its certainly interesting, for the price and performance I would go for a larger Alienware Alpha
 
Yeah I get it, I just wish intel went for 128 shading units rather adding just few % more each iteration. Like make a gaming chip family which is 10W more power hungry and put those 25W units instead of 15W ones inside NUCs.
Intel has to because AMD is superior on laptops without discrete graphics. Why own a Intel based laptop when AMD has better overall performance? The reason Iris Pro graphics were even made was because Apple was seriously considering going with AMD for their low end products. Hopefully Excavator will push AMD to competitive levels against Intel. Carrizo is said to be a 100% graphics performance increase over Kaveri. In turn Intel will have to push even harder.
As for the intel's linux drivers - those as I remember were the best linux drivers like forever, valve didn't need to push anything. nvidia had good proprietary ones and amd was like wtf i need to do to even watch a movie or webpage without tearing.
Recently it's been AMD who's been winning Linux open source driver performance. Their Catalyst drivers are on par with open source drivers, which means there's no reason to use Catalsyt in Linux. Except for 285 and 290 users. Intel in the past has had decent working drivers but the performance was still slower compared to Windows. That is until Valve fixed it.

"LunarG has been exploring the performance issues with the Intel Linux driver via their contract with Valve Software for making graphics drivers better for Linux gaming. In their research and testing they found a fundamental reason why the Intel Linux driver runs slower than the Windows driver. "

As for Nvidia, yea they have the best binary driver but open source driver is horrible. Terrible even. There's a lot of good reasons why you want open source drivers over binary.
 
Intel has to because AMD is superior on laptops without discrete graphics. Why own a Intel based laptop when AMD has better overall performance? The reason Iris Pro graphics were even made was because Apple was seriously considering going with AMD for their low end products. Hopefully Excavator will push AMD to competitive levels against Intel. Carrizo is said to be a 100% graphics performance increase over Kaveri. In turn Intel will have to push even harder.

Recently it's been AMD who's been winning Linux open source driver performance. Their Catalyst drivers are on par with open source drivers, which means there's no reason to use Catalsyt in Linux. Except for 285 and 290 users. Intel in the past has had decent working drivers but the performance was still slower compared to Windows. That is until Valve fixed it.

"LunarG has been exploring the performance issues with the Intel Linux driver via their contract with Valve Software for making graphics drivers better for Linux gaming. In their research and testing they found a fundamental reason why the Intel Linux driver runs slower than the Windows driver. "

As for Nvidia, yea they have the best binary driver but open source driver is horrible. Terrible even. There's a lot of good reasons why you want open source drivers over binary.

Intels Linux drivers aren't bad. For my purposes their biggest problem is that they lack GPU Video decode in Linux, which is a near fatal flaw, as I would otherwise use them as MythTV frontends. AMD's UVD works in linux, but requires some workarounds. Nvidia's VDPAU just works, and works perfectly in Linux. Intel seems to have absolutely no Linux video decode support at all...

AMD does have the the best of the open source drivers under Linux, no doubt, but no Linux driver beats the Nvidia closed source driver. Nothing comes even close.

As far as graphics drivers under Linux go it's something like this:

Nvidia Binary > AMD Open Source > Intel Open Source > Nvidia Open Source > AMD Binary (FireGL)

This for me is why I mostly buy Nvidia. I primarily use Linux, and the Nvidia binary driver is just that good, though I am still open to going AMD as I do most of my gaming in Windows, so I really only need rather basic desktop graphics in Linux.

I philosophically prefer open source software, but in the end I go for what works, and the Nvidia binary driver is untouchable in Linux.
 
I was amd fanboy once. I stopped rooting for them when I bought 125W FX-8320.

This cpu is powerful but again as a product for everyday user is such a crap. It overheats on its own stock cooler. I had to put big copper pot over it and lower the voltage for it to work stable. And it works on quite well rated asrock 970 PRO3. That's not a thing that user should do when he's not overclocking.

Noteworthy is that we have waited three years to get 95W octacore FX chips shipped to Europe while those were listed on the buldozer premiere and piledriver premiere.

And that's really pain to play on 125W CPU that overheat with his own boxed cooler while you could get same performance 69W Xeon which price differs by the required cooler for the FX.

Meanwhile intel has 35W chips on laptops with similar performance.

The name Excavator for the next AMD line of chips just made me giggle about whether they are preparing to dig their own grave or dig old K10 line of chips and restart their evolution.


Getting back to the topic of drivers:

AMD has quite bad proprietary drivers but they let out open source stuff quite often so the open source drivers are chasing up regularly. Sometimes they release drivers missing patches and you've got to dig in the internet to find out what is missing and attach patches yourself.

Intel has awesome open source drivers and they have specific team for the job in Israel. You are right about the performance though that increased significantly when the valve gave them debug versions of some games to work on the performance of the drivers and the source engine itself. Still it doesn't mean that the drivers or the gpu's were the issue, most likely it was the game code itself.

For the Nvidia they have the best performance on their proprietary drivers but updating system and kernel with it usually get something screwed up and you have to dig under the hood to start it running again. And for the game performance - if some extensions are missing and they should be there or they're not working as defined you need someone inside nvidia to know what to do about it with your code.

I know that stuff cause I'm following phoronix for a long time and my friend ported like 5 or 6 A+ full priced steam games to linux already.
 
Last edited:
And it works on quite well rated asrock 970 PRO3. That's not a thing that user should do when he's not overclocking.
ASRock, Quality? LOLOLOL! ASRock 970 PRO3 well rated? For me to poop on maybe?

Is this that motherboard? Firstly, that's a 4+1 phase motherboard. You couldn't even run a FX 6300 without CPU throttling, let alone a 8320. Second, there's no heatsink on the VRMs. So yea that system would have been unstable as hell. Go learn something about motherboard VRMs and the problems AMD has with them. Blame the motherboard makers for putting cheap VRMs on motherboards.

1371482073.jpeg


AMD has quite bad proprietary drivers but they let out open source stuff quite often so the open source drivers are chasing up regularly. Sometimes they release drivers missing patches and you've got to dig in the internet to find out what is missing and attach patches yourself.
AMD has developers that they pay to work on those open source drivers. So does Intel, but not Nvidia. Though Nvidia throws a bone at Nouveau guys when it comes to Tegra, which nobody cares for.
 
I've had the R2.0 version which was quite well reviewed at least in my local media. The thing is, except for high-end boards the FX-8320 and other 8-cores have problems.

On intel I can pick whatever I want price range or wattage and it works stable if the board supports it. I'm not talking about the choice of the board. I'm not talking about releasing overheating chip and letting motherboard manufacturers build something that won't work well on the boxed cooler.

And for the record I didn't say anything quality anywhere.

I tend to pick cheaper boards because those are more likely to be replaceable after the warranty expires and most likely there's going to be a new platform already when you want to replace the processor for something new. I also try to stay away from stupidly overpaid extreme overclocking motherboards like maximus formula one which booted 5 minutes to POST with 8 gigs of ram and fried power supplies because there were no standard (not overclocking) setting modes.

Finally because of all this, that you need to additionally educate yourself when picking stuff from AMD and that in the end you end up with same total amount spent on motherboard and cpu as when going for intel if you want the performance setup, its just no brainer to go for intel. I seriously doubt I know personally anyone who would go for AMD for a workstation machine.
 
Last edited:
I've had the R2.0 version which was quite well reviewed at least in my local media.
You're local media should be fired. Out of a cannon.
On intel I can pick whatever I want price range or wattage and it works stable if the board supports it. I'm not talking about the choice of the board. I'm not talking about releasing overheating chip and letting motherboard manufacturers build something that won't work well on the boxed cooler.
The 8 cores don't generally overheat any worse compaed to Intel. The 8 core 9000's though do, but nobody should ever buy those.

AMD kept the AM3+ socket going for a longer period of time to allow it to be easier to upgrade without upgrading the motherboard. Which is nice but it did allow a lot of motherboard makers to deceive consumers. That 970 Pro 3 is rated at 140W but you know that it's nowhere near capable of that. False advertising that isn't AMD's fault.

ASRock has a reputation for naming their boards with a chipset they don't have. I have a ASRock that also has the name "970" in it but doesn't actually have a 970 chipset in it, but it did have a 770. It works but 4+1 phases with no heatsink. It couldn't even handle a FX-4100 without VRM issues.

Finally because of all this, that you need to additionally educate yourself when picking stuff from AMD and that in the end you end up with same total amount spent on motherboard and cpu as when going for intel if you want the performance setup, its just no brainer to go for intel. I seriously doubt I know personally anyone who would go for AMD for a workstation machine.
You keep blaming AMD for a motherboard VRM quality issue. Call out ASRock or whomever you buy motherboards from, cause they all do it. Asus had VRM issues with their Intel X99 motherboards and nobody blamed Intel for those issues. But when Intel 8 cores started to use more power like AMD does, it started to put stress on those cheap VRMs.

It happens much less on Intel because Intel tends to release new chipsets more often than AMD. Which comes with certain standards that are imposed. You rarely find a 990FX motherboard with low quality VRMs for this reason. A 970 chipset will do just as fine as a 990FX so long as you use enough quality VRMs.

BTW Intel or AMD you shouldn't buy a motherboard with VRMs that don't have a heatsink on them. It's bad.
 
Yeah, the problem here is Asrock, not AMD.

On a side note I have a Gigabyte 990FXA UD3 with an 8+2 Phase design with heatsinks.

Got it really cheap in a Microcenter bundle with an FX-8120, $200 for both combined.

Was rock solid in my server from late 2011 until last summer, when I retired it in favor of actual server grade hardware.

Built a rig for my stepson out of it for the holidays just passed with an FX-8350 a friend wasn't using, and it's still going strong :p
 
Last edited:
I used this as an example why AMD is ridiculous. Another one would be just releasing those frakin' FX-9000's. And the simple idea that most of the boards for the AM3+ platform are flawed in use with 8-cores and you've got to waste your time researching this is also ridiculous.

You're right about the technical background of the problem but the source is still AMD's way of doing things like keeping the AM3+ socket for this whole time. I get it that its fault of the board on my end but looking for the info on what's the problem you notice that your motherboard is just the tip of the iceberg.

Another example would be whole APU desktop chips that were supposed to beat the intel in lower segments that in the end you can buy a set of better pentium + discrete graphics or i3 saving money.

In general I'm blaming AMD for being in the "middle ages". Note that since sandy bridge (2011) intel didn't really made a giant leap in performance so AMD had 4 years to catch up. We're still at the point where intel's 4 year 95W i7-2600K is by far better than fresh AMD 95W FX-8350E.
 
I used this as an example why AMD is ridiculous. Another one would be just releasing those frakin' FX-9000's. And the simple idea that most of the boards for the AM3+ platform are flawed in use with 8-cores and you've got to waste your time researching this is also ridiculous.

You're right about the technical background of the problem but the source is still AMD's way of doing things like keeping the AM3+ socket for this whole time. I get it that its fault of the board on my end but looking for the info on what's the problem you notice that your motherboard is just the tip of the iceberg.

Another example would be whole APU desktop chips that were supposed to beat the intel in lower segments that in the end you can buy a set of better pentium + discrete graphics or i3 saving money.

In general I'm blaming AMD for being in the "middle ages". Note that since sandy bridge (2011) intel didn't really made a giant leap in performance so AMD had 4 years to catch up. We're still at the point where intel's 4 year 95W i7-2600K is by far better than fresh AMD 95W FX-8350E.

AMD will likely never catch up. It's time we face that.

They simply do not have the money to spend on R&D like the giant Intel has. Intel can out spend them, out design them, out manufacture them , out everything them. AMD is simply too small, with too small cash reserves to compete.

Intel is a $160B in market cap, highly profitable company that can easily reinvest in R&D, and use it's armies of engineers to bring out next gen products.

AMD is a $2.2B in market cap company hemorrhaging money all over the place. They simply can not keep up with intel, and I don't think we will ever see their CPU's competitive again, unless someone buys them and infuses tons of cash, but that would likely never happen, due to the non-transferable x86 license.

Board compatibility - however - is not a reason to ridicule them.

IMHO, they did the right thing by keeping the socket compatibility longer, as it allows people upgrades without having to swap out the entire motherboard, as you have to with intel.

Blame the board manufacturers for lying about and mislabeling their boards, and the user for failing to do any reading and just assuming "they have the same socket, they should work". That's historically never the way it has been, only on recent intel platforms where they lock you in to a chip specific platform has that been the case.
 
+1 on that Siba :)

@Zarathustra[H] - yeah I know that simple socket check's not the way but multiple board vendors adding 8-core chips to their CPU support lists of crappy motherboards is completely different level of cpu manufacturer letting vendors break the company's trust.

Would be nice to cut it out of the broadwell topic though, sorry for offtopping so far.
 
Back
Top