TG780V3
LARGE MODEL INFERENCE AI SERVER

PLATFOORMA CUSUB, AWOODA WAAWAADEED

Wuxuu isticmaala Intel 4th/5th Gen Xeon Scalable processor platform, waxay leeyihiin horumarinta guud ee CPU compute power, PCIe speed, iyo memory bandwidth si ay ugu badiso awooda GPU.

Wuxuu ogola 8x 600W GPUs; kartida kasta wuxuu isticmaala PCIe 5.0 x16, si toos ah ugu dhaqaafo awooda computing ee xilganka.

CPU-GPU direct-connect design wuxuu garaynayaa caasimada iyo hoos u dhacnaya, si gaar ah ugu siiro farsamada data transfer dhicitaanka PCIe Switch architectures.

LA ADEEGSAN KARTA KUWA FAN IYO BLOWER

La adeegsiga waa mid weyn, waxaa ogola in la isticmaalo kartida fan-cooled (active) ama blower-style (turbine) GPU.

Wuxuu ka caawin yahay 8x four-slot fan-cooled cards ama 8x dual-slot blower-style cards.

FLEXIBLE CONFIGURATION & ON-DEMAND SELECTION

Wuxuu ka caawin yahay 11 PCIe 5.0 isla soo horay ee standard; waxaa jira mid bala ah oo kali ah ee PCIe.

1 x OCP 3.0 NIC oo aan lahayn, leh mid bala ah oo kali ah ee speed.

Wuxuu ka caawin yahay 8 x 3.5"/2.5" SAS/SATA HDDs/SSDs; kadibna wuxuu ka caawin yahay 2/4 x U.2 NVMe SSDs, taas oo kala dhaafaya astaamaha sare iyo xaafada astaamaha sare ee dhulka.

Dhibaatada, Ammaanka & Maamulka Cudurka ah

Qodobbada muhiimka ah ee nidaamka waxay leeyihiin qaabka la isku day karaa iyo la badali karaa, oo lagu daray maamulka aan ku baahan tirada, si ay u siiyaan xaqiiqda iyo ammaanka nidaamka.

Chipka maamulka cudurka ah ee la mid yahay waxay bixisa platform maamulka oo furan oo dhammaanaya baruuraha kale ee ah IPMI 2.0, Redfish, iyo SNMP.

Wuxuu ka caawin yahay mid bala ah oo kali ah ee caawina oo kale ah oo leh KVM, Virtual Media, monitoring ee xaaladda arkhaynta muhimka ah, dhibaatooyinka dhaafashada, iyo kale, taas oo helisa caawina dhan ee remote ee nidaamka iyadoo leh qaadka caawina.

Beijing Puhua Haotian Technology Co.