News Intel GPUs - Battlemage officially announced, evidently not cancelled

[igor_kavinski]

I think their engineers have gotten into the habit of debugging silicon and getting revised steppings (keeping their IFS brethren busy) rather than finding and fixing issues in the initial design phase. That would explain the 12 Sapphire Rapids steppings and ARC might have had at least half that many steppings.

Who knows? Maybe their validation woes are still ongoing ever since BK fired the validation team. They lost their senior validation engineers and maybe still having trouble finding competent people to fill those roles.

 

[DAPUNISHER]

IMO, at this end of the market, people are probably looking at upgrading old PCs which has suddenly become an issue for me. Since a friend gave me an old HP with Ivy Bridge. That's actually a CPU upgrade over my Core 2 Quad.

But it has proprietary MB and PSU, and no GPU power plugs, so limited to 75W.

Researching it looks like the best card you can buy in this situation is the GTX 1650. Even if the A380 was 75W, it needs REBAR, and doesn't touch a GTX 1650 for performance, RX 6400/6500 are crippled by their 4 channel PCIe.

At playable settings for both cards, the 6500XT at PCIE 3.0 still beats the 1650 in most games. Note: I ignore the derp settings reviewers use, that make all the budget cards look bad.

My bad, I see he has no power connector. Can't pull the few extra watts from SATA? Some people pearl clutch about it, but it works fine if you are not pushing too much through it.

 

[DAPUNISHER]

Something is off with the RX6500 XT result in Doom Eternal.

RX6400= 52 fps
A380= 56 fps
RX 6500XT= 98 fps

Difference between 6400 and 6500 XT is way too big. Here are results from techpowerup which seems more plausible: https://www.techpowerup.com/review/msi-radeon-rx-6400-aero-itx/14.html

HUB only tested at ultra, doesn't TPU test max settings?

 

[mikk]

HUB only tested at ultra, doesn't TPU test max settings?

Doesn't matter if it's max or ultra, it's about the differences, something is wrong with the RX 6500 result from Hardware Unboxed.

 

[Stuka87]

Researching it looks like the best card you can buy in this situation is the GTX 1650. Even if the A380 was 75W, it needs REBAR, and doesn't touch a GTX 1650 for performance, RX 6400/6500 are crippled by their 4 channel PCIe.

The 4 channel PCIe is really only an issue in games that want more than 4GB of RAM. The reviewers that like to hate on the 6400/6500, are the ones that purposely ran way higher settings than the cards would realistically be run at.

The 1650 doesn't have the PCIe issue. But it does have the nVidia driver CPU overhead issue. So I/O will be faster, but the slower CPU you have will hurt it. Though it would require some testing to see which limitation is the larger issue.

 

[DAPUNISHER]

Doesn't matter if it's max or ultra, it's about the differences, something is wrong with the RX 6500 result from Hardware Unboxed.

You might be right. But knowing Steve, it is because his test was in an easier area.

 

[Heartbreaker]

At playable settings for both cards, the 6500XT at PCIE 3.0 still beats the 1650 in most games. Note: I ignore the derp settings reviewers use, that make all the budget cards look bad.

My bad, I see he has no power connector. Can't pull the few extra watts from SATA? Some people pearl clutch about it, but it works fine if you are not pushing too much through it.

6500xt uses more power and is out just for that. I'm not about to push the PSU or it's connectors on a 10 year old HP PSU. I'm a little nervous taxing it with a discrete GPU at all.

Performance I see is on average identical between 6500XT on PCIe 3.0, and the Chipset in the HP only supports PCIe 2.0, so that would clearly make it much worse.

To top it off, 6500xt costs more here in Canada.

 

[Insert_Nickname]

Says he'd take a chance on it v. 6400 because he doesn't play old games and it has hardware media support.

The media capabilities are literally the only reason I'd get one. The A310 will be an even better fit for IGP-less non-gaming Ryzens.

Being able to run a few games is just a bonus.

The 4 channel PCIe is really only an issue in games that want more than 4GB of RAM. The reviewers that like to hate on the 6400/6500, are the ones that purposely ran way higher settings than the cards would realistically be run at.

I'd really like to see a comparison between the regular 4GB and the new 8GB variant. That should put the memory debate to rest.

Anyone seen that yet?

 

[LightningZ71]

6500xt uses more power and is out just for that. I'm not about to push the PSU or it's connectors on a 10 year old HP PSU. I'm a little nervous taxing it with a discrete GPU at all.

Performance I see is on average identical between 6500XT on PCIe 3.0, and the Chipset in the HP only supports PCIe 2.0, so that would clearly make it much worse.

To top it off, 6500xt costs more here in Canada.

Ugh, are you sure that you even have the power budget for a 75 watt card in the PCIe slot? While the soec calls for it, some of those lower end proprietary board and PSU combinations could only handle 25, 50 or 60 watts on the slot max, not to mention if the PSU even has enough wattage for it.

 

[DAPUNISHER]

6500xt uses more power and is out just for that. I'm not about to push the PSU or it's connectors on a 10 year old HP PSU. I'm a little nervous taxing it with a discrete GPU at all.

Performance I see is on average identical between 6500XT on PCIe 3.0, and the Chipset in the HP only supports PCIe 2.0, so that would clearly make it much worse.

To top it off, 6500xt costs more here in Canada.

Fair enough. Good luck with the build, hopefully the PSU takes the strain.

 

[DAPUNISHER]

The media capabilities are literally the only reason I'd get one. The A310 will be an even better fit for IGP-less non-gaming Ryzens.

Being able to run a few games is just a bonus.



I'd really like to see a comparison between the regular 4GB and the new 8GB variant. That should put the memory debate to rest.

Anyone seen that yet?

I agree that the excellent media features are enough for it to be compelling for HTPC use.

I haven't seen any testing of the 6500XT 8GB version yet. I liked to see it show up under $200 and the 6400 and 6500 get a $30+ price cut. Probably not, since the competition in the space is completely underwhelming.

 

[jpiniero]

I haven't seen any testing of the 6500XT 8GB version yet. I liked to see it show up under $200 and the 6400 and 6500 get a $30+ price cut. Probably not, since the competition in the space is completely underwhelming.

Unlikely since you have to figure that AMD completely ended production of Navi 24 once they realized they weren't going to get any mobile OEM deals out of it.

Newegg has it for $229. That sounds about right. Course it's a bad deal compared to the 6600.

 

[DAPUNISHER]

Ugh, are you sure that you even have the power budget for a 75 watt card in the PCIe slot? While the soec calls for it, some of those lower end proprietary board and PSU combinations could only handle 25, 50 or 60 watts on the slot max,

That's a good point. My 6400 has yet to peak at more than 43W gaming. Aussie Steve can bash all he wants; haters gonna hate. But my overclock 5800x + 3060ti is sitting here unplugged while I play games like the Fallout series, AC Black Flag, Shadow of Mordor, and GTA5 on my 5600@4.65GHz + 6400.

 

[DAPUNISHER]

Unlikely since you have to figure that AMD completely ended production of Navi 24 once they realized they weren't going to get any mobile OEM deals out of it.

Newegg has it for $229. That sounds about right. Course it's a bad deal compared to the 6600.

Yup. Going to be another dust collector. While Intel and Nvidia are apparently not going to shake the $200 and under market up, used cards are starting to.

 

[Heartbreaker]

Ugh, are you sure that you even have the power budget for a 75 watt card in the PCIe slot? While the soec calls for it, some of those lower end proprietary board and PSU combinations could only handle 25, 50 or 60 watts on the slot max, not to mention if the PSU even has enough wattage for it.

I've seen several other reports of people successfully putting 75W GPU cards in these systems. This is the tower version with 320W PSU. Though I am a bit nervous just the same, and have not pulled the trigger on a GPU.

 

[Heartbreaker]

GN dumps on the drivers big time:

 

[mikk]

GN dumps on the drivers big time:

Arc control panel included in the driver is really outdated, they should have tested the newest which is available here: https://www.intel.com/content/www/us/en/download/727323/intel-arc-control.html

Of course it's mind-boggling that Intel doesn't include the newest control panel in the driver, what the hell they are doing. As I said earlier they have to remove the responsible person for this mess, it's Lisa Pearce. 10+ years head of the drivers team and look in which mess this is. It's like they don't care.

 

[Leeea]

GN dumps on the drivers big time:

Indicated his overclocking attempts resulted in a 2% performance increase.

Very different from other reports.

Steve also seems profoundly frustrated after attempting to use it for two weeks.

 

[moonbogg]

During the mining boom intel could have charged $800 for this thing, even if it can't mine, and people would have still panic bought millions of them for some odd reason. Don't you dare disagree either because I know you saw people selling PICTURES of RTX 3080s for $1000 on ebay during that time, and they were SELLING.

 

[maddie]

GN dumps on the drivers big time:

I honestly can't remember a harsher review. This is horrible. Who will be the lab rats? I don't think I would want that in my machine even for free.

 

[HurleyBird]

GN dumps on the drivers big time:

Wow. It almost seems like, somehow, Intel's GPU driver team is comprised of terrible developers. Not being able to properly optimize the driver is one thing, putting out actual terrible software is another. It's like someone just hired all their friends and family members. No competent developer should build something with such a broken UI, even absent a QA department. What happened?

 

[Stuka87]

As somebody that works in software develop, and has a ton of experience on the testing side, this really hurt to watch. Somebody had to sign off on this software. It was hopefully under protest, but this is just so bad. Its one thing for a an advanced feature to maybe not work as well as they hoped. Its another thing for it to not even install correctly. Or for the UI to just be broken. This is just mind boggling.

 

[HurleyBird]

Somebody had to sign off on this software. It was hopefully under protest

Speaking as a developer who has built his fair share of complex UIs, it is clear there was a total failure in QA and management. At the same time, no competent developer should build something this broken in the first place, regardless of oversight. I doubt there are any unsung heroes here. What we're seeing is pure incompetence from top to bottom.

 

[linkgoron]

I can't believe that I'm saying this - but I don't think that it's correct blaming QA or the devs. I would have to assume that management pushed the dev and QA teams to release a full list of features no matter what, and that's what they did. I assume that the schedule was extremely harsh and QA and devs were raising all of the alarms, but management just told the to ship whatever even if it's alpha or beta quality at best.

There's no way the devs/QA don't know the state of what they were releasing - it has to be the management that's telling them to ship whatever no matter what.

 

[coercitiv]

I can't believe that I'm saying this - but I don't think that it's correct blaming QA or the devs.

I tend to agree, it looks a lot more like the management had no clear idea of what they were doing, and it was mission impossible for the devs. The worst part is that the software foundation is likely in a precarious state as a result of this chaos, and buggy or undesirable behaviour will be inherited for quite some time starting from now.